# Audience documentation --- # What Audience does URL: https://docs.audience.so/start-here/what-audience-does > The problem Audience solves for a marketing team, what Scout does about it, and which decisions stay with you. Audience is a content strategy and production system for in-house B2B marketing teams. It connects to your website and your search data, finds the topics worth writing about, explains why each one matters, and carries the ones you approve through research, outline, draft and review. Scout is the agent that does the reading, judging and writing. You are the one who decides. ![Home, the weekly briefing](/docs/screens/home.png) _Home answers one question first: what is worth moving forward this week._ ## The problem Most teams publish on guesswork. Someone has a hunch about a topic, a writer produces two thousand words, and the team finds out three months later whether anyone searched for it. The queue of "what should we write next" is the bottleneck, and it is usually filled by whoever argued loudest in the planning meeting. Agencies do not fix this. They scale words, not judgement. A brief arrives, an article comes back, and the strategic question of whether the topic fits what you sell and who you sell it to is answered by nobody, or answered once and never revisited as the business moves. The result is a blog full of pieces that are individually fine and collectively aimless: articles that rank for searches your buyers never make, that compete with your own pages, or that name a market you left last year. ## What Audience does about it Scout starts by reading. It crawls your public pages to learn what you sell and who you sell it to, and it writes that down as a company profile you can see and edit. It reads your search data to find what people already type to reach you, which of your pages are slipping, and which are nearly on page one. It looks at who ranks around you and what they hold that you do not. From all of that it builds a queue of opportunities. Each one is a topic with a score out of 100, a proposed title, an angle, and the reasoning behind it. The reasoning is not a summary: every claim points at the measurement it came from, so you can read Scout's thinking and disagree with it before committing a writer. Every topic is judged against your company profile before it reaches you. A topic with a huge search volume that has nothing to do with your business is set aside or flagged, not scored high because a lot of people search for it. Fit with what you sell is the heaviest part of the score. When you approve a topic, Scout researches it, proposes an outline for you to approve, writes a draft, and then reviews its own draft against a standard editorial rubric and against the rules in your playbook. You get a draft with a quality score, a list of findings, and the sources behind every claim. ## What stays with you Audience is built around a principle the product states on its Home page: approve the thinking before the writing. The decisions that matter are yours, and the product will not take them for you. - **Which topics move forward.** Scout proposes; you press **Move forward** or **Not worth it**. A topic that fails a guardrail is flagged and waits for you to say **Do it anyway**, or to leave it. - **Whether the outline is right.** Nothing is drafted until you have approved the outline, and what you save becomes the version Scout writes from. - **Whether the draft goes out.** Scout reports what it found and why. It does not change the draft on its own. You approve, or you send it back with notes. - **What Scout believes about your company.** The profile it built from your website is yours to edit, and when Scout re-reads your site it proposes changes rather than making them. - **What it will never propose.** Reject a topic, and you can keep topics like it out from then on. Scout learns from those decisions. Every rejection, every edit, every dismissed finding is read, and the patterns come back as proposed rules you can accept or turn down. ## What Audience is not It is not an auto-publisher. In this release an approved article leaves Audience as a Markdown download; you put it live yourself, in your own CMS, when you choose to. Nothing Scout writes becomes public without a person deciding it should. It is not a quota machine. There is no target number of articles a month. A week with nothing worth moving forward is reported as exactly that. It is not a dashboard. The numbers are there, but the product opens on what to do, not on how things are going. See [How Scout works](/start-here/how-scout-works) for the loop end to end, and [Who does what](/start-here/who-does-what) for which decisions each role can make. --- # How Scout works URL: https://docs.audience.so/start-here/how-scout-works > The loop Scout runs, from reading your site to a reviewed draft you can download, and the points where you decide. Scout runs one loop, over and over: find topics, judge them, hand you the decision, then research, outline, draft and review the ones you approve. Every stage leaves a record you can read, and the loop stops at a person wherever a decision matters. This page walks it end to end. ## 1. Scout reads your site and your search data Setup gives Scout a website address. It crawls your public pages and writes a company profile: what you sell, who you sell it to, the markets you are in, the topics you own, and the claims it should never make. You can read and edit every field of it in the [Playbook](/playbook/your-company-profile). If you connect Search Console, Scout also reads what people already type to reach you, which pages are slipping, and which queries sit just off page one. That is the strongest signal it has, and it is free to connect. See [Connect Search Console](/setup/connect-search-console). ## 2. Scout finds candidate topics Candidates come from several signals at once: queries you nearly rank for, pages losing ground, competitors' pages you have no answer to, expansions around what you sell, and the pages of yours that already work. Each signal is a lane, and the queue records which lane surfaced each topic. Where a signal is unavailable, the run says so rather than pretending it looked. ## 3. Scout sets aside what is plainly off-market Search volume knows nothing about your business. Before any topic is judged, Scout asks one question of each candidate: would this company have a reason to write about it for the people it sells to? Only a confident no is set aside. The queue tells you how many were set aside and shows the most searched of them, so you can check Scout's judgement. See [Topics Scout set aside](/opportunities/topics-scout-set-aside). ## 4. Scout judges fit against your company profile What survives is judged three ways: does the topic sit in the market you sell into, would an article on it lead to something you sell, and is the person searching it one of your buyers. Fit is the heaviest part of the score, and the judgement is written out in plain sentences on each opportunity. If you have told Scout where the business is heading, that direction outranks anything it read on your website. Change the profile or the direction and every open topic is judged again. See [Direction and re-judging](/opportunities/direction-and-re-judging). ## 5. Scout scores and applies your guardrails Each topic gets a score out of 100 built from eight components, each with its own explanation and evidence. A component Scout could not measure is marked "not measured" and its weight moves to the others; it is never counted as a zero. Then the guardrails: minimum search volume, maximum keyword difficulty, minimum score, minimum fit with the business. A topic that fails one is **Flagged**, not deleted. It waits in the flagged pile with the reason attached, one click from the queue. See [The score](/opportunities/the-score) and [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). ![The opportunity queue, best first](/docs/screens/opportunities.png) _The queue is ordered by score and nothing else. Each row says which signal found it and why Scout thinks it is worth doing._ ## 6. You decide This is the first checkpoint, and the product's principle is stated on Home: approve the thinking before the writing. You read the reasoning, and you press one of three buttons. - **Move forward** approves the topic and starts research. - **Not worth it** rejects it. An optional reason teaches Scout what you do not want, and you can keep topics like it out from then on. - **Do it anyway** overrides a guardrail on a flagged topic, with your reason recorded. You can also hand Scout a topic of your own with **Research a topic**. It goes through the same judgement and scoring as everything else. See [Deciding](/opportunities/deciding) and [Research a topic](/opportunities/research-a-topic). ## 7. Scout researches and proposes an outline An approved topic becomes a content item. Scout reads what ranks for the search, what searchers want, what nothing ranking covers, and the facts it can source. It writes an outline and stops. You read the outline, edit it if you want, and press **Approve outline & start drafting** or **Write it again** with a note. What you save is the version Scout drafts from. See [Topic and outline review](/content/topic-and-outline-review). ## 8. Scout drafts and reviews its own work Scout writes the draft from the approved outline and the research, then reviews it against a standard editorial rubric and against every rule in your playbook. The result is a quality score out of 100, a routing of **Ready for review**, **Needs revision** or **Blocked**, and a list of findings. Every claim is traced back to a source Scout read; one that cannot be traced is a finding. ![Draft review, with the findings rail](/docs/screens/draft-review.png) _The article on the left, what Scout found on the right, and two decisions at the bottom._ ## 9. You approve, or send it back You read the draft and decide. Each finding can be applied, dismissed with a reason, or ticked to go back with the draft. **Send back** returns it to Scout with your notes; **Approve & move forward** marks the version approved. Scout does not change a draft on its own. See [Draft review](/content/draft-review) and [Flags, comments and sources](/content/flags-comments-and-sources). ## 10. The article leaves as Markdown An approved article is yours to publish. In this release it leaves Audience as a Markdown download, and you put it live in your own CMS. See [Publishing and Markdown](/content/publishing-and-markdown). ## What keeps running Scores on open topics refresh daily against new search data. The full rebuild, which reads the market again, runs on the cadence you set for the workspace. Once a week Scout looks at your edits and proposes rules it thinks it has learned, which you accept or turn down. Nothing that spends against your budget happens without a button or a schedule you can see. See [What runs when](/automation/what-runs-when). --- # Key concepts URL: https://docs.audience.so/start-here/key-concepts > The dozen ideas Audience is built on, each in a paragraph, with a link to the page that covers it in depth. Audience uses a small vocabulary consistently. These are the ideas you will meet on every screen, in the order you tend to meet them. The [glossary](/reference/glossary) has the shorter, alphabetical version. ## Workspace A workspace is one company's space: its site, its profile, its playbook, its queue, its people. Every address in Audience starts with the workspace. You can belong to several and switch between them from the sidebar. See [Workspace and members](/settings/workspace-and-members). ## Site The website Scout reads and measures. You give it in the first setup step, and everything else follows from it: the crawl, the search data, the competitors, the questions asked of AI assistants. A workspace has one site. See [Your website](/setup/your-website). ## Company profile What Scout believes about your business: a summary, your positioning, your market, your products and audiences, the topics you own, and the claims it should never make. Scout writes it from your website; you can edit any field. Every topic is judged against it, so the profile is where a wrong kind of queue is fixed. See [Your company profile](/playbook/your-company-profile). ## Direction The one part of the profile a person states rather than Scout infers: what you are focusing on and what you are moving away from. It outranks anything Scout read on your website, because a website says what you sold last year. Setting it re-judges every open topic. See [Direction](/playbook/direction). ## Playbook The rules Scout writes to, grouped by category, each with a severity and a history of its wording. Every draft is graded against them. Rules come from a standard editorial rubric, from documents you upload, and from patterns Scout notices in your edits and proposes back to you. See [Rules](/playbook/rules). ## Opportunity A topic Scout thinks is worth writing about, with a score, a proposed title, an angle, the keywords it groups, the signal that found it and the reasoning behind all of it. Opportunities sit in a queue ordered by score, waiting for a decision. See [Reading the queue](/opportunities/reading-the-queue). ## Score A number out of 100 built from eight components: fit with your business, what your own search data says, how many people search, whether you can rank, what competitors own, whether it collides with your pages, where demand is heading, and outcome and AI visibility. Each component explains itself, and one Scout could not measure says so rather than counting as zero. See [The score](/opportunities/the-score). ## Guardrails Four thresholds a topic must clear to reach the queue: minimum search volume, maximum keyword difficulty, minimum score, and minimum fit with the business. Each is checked and reported on its own. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). ## Flagged A topic that failed a guardrail. It is kept, not deleted: out of the queue and out of the bin, with the failure and the numbers attached. You can put it in the queue anyway with **Do it anyway**. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). ## Content item An approved opportunity being carried through production. It moves through research, outline review, drafting, draft review and approval. Three of those stages wait on a person, and the pipeline says which. See [The pipeline](/content/the-pipeline). ## Versions Nothing is overwritten. Every outline and every draft is a version, marked as written by Scout or by the person who edited it, with a summary of what changed. One version is marked approved. See [Versions and approval](/content/versions-and-approval). ## Review Scout's reading of its own draft: a quality score out of 100, a routing of Ready for review, Needs revision or Blocked, and a list of findings. Findings are decided one at a time. Dismissing one with a reason teaches Scout to stop raising it. See [Draft review](/content/draft-review). ## Prompt panel The set of questions your buyers ask AI assistants, which Scout asks every day and reads for whether you came up, where, and how you were described. You choose the questions and confirm the panel; nothing is asked or paid for until you do. See [Build your prompt panel](/ai-visibility/build-your-prompt-panel). ## Insights Patterns Scout finds in your own numbers once there is enough published work to compare: cohorts that convert better, pages losing ground, stages that keep costing time. Each comes with its evidence and a recommended next step, and you accept or dismiss it. See [Reports overview](/reports/overview). ## Activity Everything Scout is doing and has done, in the open: every run, every agent call, how long it took and what it consulted. The sidebar shows a live indicator whenever Scout is working, and the Activity page has the full record. See [Activity](/automation/activity). --- # Who does what URL: https://docs.audience.so/start-here/who-does-what > The four workspace roles, what each can decide, and the two rules that shape them: everyone can read, and approving is not publishing. A workspace has four roles: Owner, Admin, Editor and Viewer. Two rules explain most of the table below. First, every member can read everything in a workspace they belong to; there is no role that sees less. Second, the decisions are separated by weight: an editor can approve a topic and a draft, but the settings, the integrations and the act of putting something live belong to owners and admins. Where a screen cannot offer you an action, it says so rather than hiding the button. You will see lines like "Only editors and above can start research" and "Only owners and admins can change where content is published". ## The roles | What you can do | Owner | Admin | Editor | Viewer | | --------------------------------------------------------------------------------------------------- | :---: | :---: | :----: | :----: | | See every screen in the workspace | Yes | Yes | Yes | Yes | | Comment on a draft | Yes | Yes | Yes | Yes | | Decide an opportunity: **Move forward**, **Not worth it**, **Do it anyway**, **Find another angle** | Yes | Yes | Yes | | | **Research a topic** (spends against the workspace budget) | Yes | Yes | Yes | | | Approve or send back an outline or a draft; apply or dismiss findings; vouch for a source | Yes | Yes | Yes | | | Add or remove a topic exclusion | Yes | Yes | Yes | | | Accept or dismiss a report insight | Yes | Yes | | | | Edit the playbook rules and decide learned rules | Yes | Yes | | | | Edit the company profile, set direction, accept a re-read | Yes | Yes | | | | Configure and run the AI visibility prompt panel; measure the market | Yes | Yes | | | | Connect, re-check or disconnect integrations; choose a property | Yes | Yes | | | | Manage publishing destinations | Yes | Yes | | | | Run setup and finish the build | Yes | Yes | | | | Rename the workspace, set the rebuild cadence, invite people, change roles, remove members | Yes | Yes | | | | Delete the workspace | Yes | | | | ## What the shape means in practice **Viewers read and comment.** A viewer sees the queue, the drafts, the reports and the playbook exactly as everyone else does, and can leave a comment on any passage of a draft. They cannot move anything. **Editors run the content work.** An editor decides topics, starts research, approves outlines and drafts, and can keep a theme out of the queue. Excluding a topic follows the same permission as deciding one, because it is a decision about the queue, not about the playbook. **Editors do not publish.** An editor can approve a draft; putting it live is a separate act that belongs to owners and admins. In this release that act is downloading the Markdown and publishing it yourself, and the Publishing settings page says so. The separation is deliberate, not an oversight. **Owners and admins run the workspace.** Setup, integrations, the company profile, the playbook, the prompt panel and the members list are theirs. The difference between the two is small: only an owner can delete the workspace. Owners cannot be demoted or removed from the members screen, and cannot leave. ## Where roles are set Roles are assigned when you invite someone and can be changed on the workspace's members list. The invite form offers Admin, Editor and Viewer; ownership comes from creating the workspace. See [Invite your team](/setup/invite-your-team) and [Workspace and members](/settings/workspace-and-members). --- # Set up your workspace URL: https://docs.audience.so/setup/set-up-your-workspace > The five setup steps, what you need before you start, and how long it takes to go from a website address to a full queue of opportunities. Setting up takes about a minute of your time and, after that, a few minutes of Scout's. You give it a website address, connect the search data you already have, optionally hand it anything already written down, confirm who you compete with, and then let it build. When it finishes, Home tells you what it learned and the queue is ready to review. ![Step 1 of setup, asking for the website address](/docs/screens/onboarding-website.png) _Every step carries the same frame: which workspace is being set up, its domain, and where you are in the five steps._ ## Before you start You need three things, and only the first is required. - **An invitation.** Accounts are created by invitation. A workspace owner or admin invites you, you open the link in the email, and you create your account from there. See [Invite your team](/setup/invite-your-team). - **Your website address.** The one your customers visit, for example `resemble.ai`. Scout works out the rest. - **Access to Google Search Console** for that site, ideally as the account that owns the property. This is what step 2 connects. Google Analytics is optional. ## The five steps Each step is a page with its own address, so you can leave part-way through and come back from a bookmark, a second device, or the Google consent screen. The heading above every step reads **Setting up {workspace} · {domain} · Step N of 5**, and optional steps are marked **Optional** in the stepper with a real skip button rather than a hidden link. | Step | What you do | How long | | ---------------------------------------------------------- | ------------------------------------------------------------------------------- | ----------------------------- | | 1. [Your website](/setup/your-website) | Give Scout the address. It starts reading straight away. | Under a minute | | 2. [Connect Search Console](/setup/connect-search-console) | Sign in with Google and pick the property. Strongly recommended. | A minute | | 3. [Add your knowledge](/setup/add-your-knowledge) | Upload brand guidelines, positioning or a favourite article. Optional. | As long as you like | | 4. [Confirm competitors](/setup/confirm-competitors) | Check the list Scout found. Drop anyone who does not belong, add anyone missed. | A minute | | 5. [The build](/setup/the-build) | Scout puts the workspace together. You can leave. | A few minutes of Scout's time | Reading your site and pulling your search history carry on in the background from the moment you start them. You never have to wait for a step to finish before moving to the next. ## Who does the setting up Setting a workspace up is left to its owners and admins. Anyone else in the workspace sees the same steps with a note: **Setting this workspace up is left to its owners and admins. You can follow along here.** They can watch progress but cannot change anything. Roles are explained in [Who does what](/start-here/who-does-what). ## Coming back to an unfinished setup Opening the workspace's setup address always lands you on the step that is still waiting. If Scout is already building, Home shows a progress panel with the current step and what is still coming, and fills itself in as each piece lands. Nothing stops when you close the tab. ## What happens next When the build finishes, the last step reads **Here is what Scout learned about {domain}.** with the counts of what it found, and one button: **Review what Scout found**. That takes you to the [opportunity queue](/opportunities/reading-the-queue), which is where the work starts. --- # Your website URL: https://docs.audience.so/setup/your-website > Step 1 of setup. You give Scout one address, and it reads your public pages to learn what you sell, who you sell it to, and what you already rank for. Step 1 asks for one thing: the address your customers visit. Everything Scout knows about your company starts here, so it reads the site before it says anything. The page opens with the question **Let's find what moves you forward.** and a single field labelled **Website address**, with the placeholder `example.com`. The helper text says it plainly: _The address your customers visit. Scout works out the rest._ You do not need to add `https://`, a `www`, or a path. ## What happens when you press "Start reading my site" Two things start immediately, and both keep running in the background while you carry on with setup. - **A read of your public pages.** Scout crawls the site to learn what you sell, who you sell it to, how you describe yourselves, and what you have already written. This becomes the first version of your [company profile](/playbook/your-company-profile) and the starting point for your [playbook](/playbook/rules). - **A market baseline.** Scout looks up what the domain already ranks for and who else ranks for the same searches. That is where the first list of [competitors](/setup/confirm-competitors) comes from, and it is the "before" picture your [Reports](/reports/overview) measure against. The same screen turns into a progress view showing both tasks. You can watch, or you can move on. The footer says so: **Reading your site keeps going in the background. You do not have to wait for it.** ## If you typed the wrong address Below the progress view is a disclosure labelled **Wrong address?**. Open it to enter the address again. Scout starts over with the new one. ## Moving on The footer button reads **Next — connect your search data**. Step 2 is where you [connect Search Console](/setup/connect-search-console), which is the single strongest signal Scout has. ## Notes - One workspace is about one website. The workspace switcher shows the domain under the workspace name once a site exists. - Only owners and admins can add a website. Everyone else sees: _Nobody has added a website yet. An owner or admin of this workspace needs to do that before Scout can start._ - Scout reads public pages only. Anything your website does not say, such as who you never write for or claims you cannot make, belongs in [step 3](/setup/add-your-knowledge) or in the [company profile](/playbook/your-company-profile) later. --- # Connect Search Console URL: https://docs.audience.so/setup/connect-search-console > Step 2 of setup and the Integrations settings page. Search Console is the single strongest signal Scout has, what it unlocks, and what Scout does without it. Search Console tells Scout which queries already bring people to your site and which of your pages are slipping. It is the single strongest signal Scout has, and it costs nothing to connect. The setup card marks it **Strongly recommended** for that reason. The step opens with **Connect what people already search.** and two cards: Search Console, and [Google Analytics](/setup/connect-google-analytics), which is marked **Optional**. ![The Integrations settings page with the Search Console and Analytics cards](/docs/screens/settings-integrations.png) _The same two cards appear in Settings after setup, with Sync now, Reconnect and Disconnect._ ## Connecting 1. Press the connect button on the Search Console card. Google's own consent screen opens. 2. Sign in with the Google account that has access to the property for your site, and allow read access. 3. Google sends you back to Audience. Scout matches your site's domain against the properties on the account. If exactly one property matches, Scout picks it and the card reads **Reading {property}.** If the match is genuinely ambiguous, for example a domain property and a URL-prefix property for the same site, a property picker appears and you choose. Scout starts pulling your history in the background as soon as the property is set. The footer notes: _Scout is pulling your history in the background._ ## What it unlocks | Where | With Search Console | | --------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- | | Every [opportunity score](/opportunities/the-score) | The "what your own search data says" component is measured rather than marked not measured | | The [opportunity queue](/opportunities/reading-the-queue) | Topics you are close to ranking for, pages that are slipping, and queries where people see you but do not click | | [Reports](/reports/overview) | Organic clicks, impressions, click-through rate and position, and the whole Overview tab | | An opportunity's "Against what you already publish" panel | Search Console overlap with your existing pages | | [Content performance](/reports/content-performance) | How each page Scout published is doing, week by week | ## Without it Scout still works. It reasons from market data instead, and it says so wherever the gap shows rather than pretending. - The first-party component of every score is marked **not measured** and its weight moves to what Scout could measure. The score explanation records _Capped — Search Console is not connected._ - Home's clicks card reads **Connect Search Console**. - Reports Overview says **Nothing to report yet.** until the first sync lands. - The cannibalization panel on an opportunity says **not connected** for overlap. If you skip the step, the footer reads **Continue without it**, and a note follows: _You can connect Search Console any time from Settings. Scout will work from market data until you do._ ## Keeping it connected Open **Integrations** from the workspace menu at any time. - **Sync now** re-checks the connection and pulls what is new. - **Continue connecting** appears if a consent flow was started but never finished. - **Reconnect** appears if the grant expired, failed or was revoked. - **Disconnect** stops Scout reading new data. Disconnecting does not delete anything. As the page says: _The history it has already pulled stays where it is, so reconnecting the same property picks up where it left off._ ## What happens next Once connected, recent search data refreshes every morning, and Scout re-scores every open opportunity against it. See [What runs when](/automation/what-runs-when). The next step is [Add your knowledge](/setup/add-your-knowledge), which is optional. --- # Connect Google Analytics URL: https://docs.audience.so/setup/connect-google-analytics > Optional in setup and in Settings. Analytics lets Scout report what happened after the click, so Reports show signups and revenue instead of only traffic. Search Console tells Scout how people found you. Analytics tells it what happened next. The card in step 2 is marked **Optional**, and its purpose line says what it is for: _What happens after the click, so Scout can report signups and revenue instead of traffic._ The product's phrase for this is **Measure outcomes, not output.** Clicks are the means; conversions are the end. Without Analytics, Scout can only report the means. ## Connecting The flow is the same as [Search Console](/setup/connect-search-console): press connect, sign in with Google, allow read access, and come back. Scout then needs to know which Analytics property to read. If the property is unambiguous, Scout picks it. If it is not, you choose from a list. Until you have chosen, the connection is in a distinct state Audience treats honestly rather than as failure: - Home's outcome card reads **Choose a property** with a link, **Choose which property to read**. - The build task in step 5 shows **Waiting on you**. Telling someone to connect Analytics when they already have is the fastest way to make them distrust a screen, so Audience never does that. Connected-but-unconfigured is its own state, with its own fix. ## What it unlocks | Where | With Analytics | | --------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Home | The **Business outcome** card, labelled from your property, for example _Signups from organic_ | | [Reports Overview](/reports/overview) | The **Outcomes** section: Sessions, **Key events** (the conversions your Analytics property counts), and **Influenced revenue** from pages Scout can attribute | | [Content performance](/reports/content-performance) | A **Key events** column beside clicks and impressions, so a page's conversions sit next to its traffic | | What Scout notices | Findings about which content converts, not only which content attracts | Revenue appears only when Analytics actually records it. If your property has no revenue metric, Reports say **No revenue metric configured** rather than showing a zero. ## Without it Home's fourth card carries the invitation: _Measure outcomes, not output. Connect Google Analytics and Scout reports on signups or revenue instead of clicks._ Reports Overview shows **Outcomes are not measured yet.** in place of the outcomes panel, and the Key events column is absent from Content performance. Nothing else changes; the opportunity queue and the content pipeline do not depend on Analytics. ## Keeping it connected Analytics sits beside Search Console on the **Integrations** page, with the same controls: **Sync now**, **Reconnect** when a grant has expired, and **Disconnect**. Disconnecting stops new data; the history already pulled stays where it is, and reconnecting the same property picks up where it left off. ## Related - [Reports Overview](/reports/overview) for how outcomes are shown once they exist. - [What runs when](/automation/what-runs-when) for the daily Analytics sync. --- # Add your knowledge URL: https://docs.audience.so/setup/add-your-knowledge > Step 3 of setup, and optional. Brand guidelines, positioning, a content SOP or an article you were proud of become playbook rules Scout checks every draft against. Your website says what you sell. It rarely says how you write, who you never write for, or which claims legal will not sign off. Step 3 is where you tell Scout those things, by giving it what is already written down. The step opens with **Anything already written down?** and explains itself: _Brand guidelines, positioning, your content SOP, an article you were happy with. Scout turns these into playbook rules it checks every draft against._ ![Step 3 of setup, the knowledge step with its empty state](/docs/screens/onboarding-knowledge.png) _Before anything is added, the page says "Nothing yet, and that is fine"._ ## What you can add - **Files.** PDF, Word, Markdown, plain text, HTML and spreadsheets, up to 50 MB each. - **Links.** Public document links, one per field, with **Add another link** to add more. Press **Add to the playbook** to submit. Everything you have given Scout appears in the list headed **What Scout has been given**, each with a **Remove** control. A workspace holds up to 50 documents. At the limit the page says: _This workspace is holding its limit of 50 documents. Remove one to add another._ ## What Scout does with it Scout reads each document and extracts the rules it finds: the voice you write in, the words you avoid, the audiences you serve and the ones you do not, the claims you make and the ones you must not. Each becomes a rule in your [playbook](/playbook/rules), with its provenance attached, so you can see which document it came from and the passage it was read from. Rules from your documents sit alongside the standard editorial rubric every workspace starts with, and every draft is graded against both. Nothing here is a lock. A rule can be reworded, turned off or deleted from the Playbook at any time. ## Skipping it The step is optional and says so. The empty state is titled **Nothing yet, and that is fine**, and the footer note reads: _This step is optional. Scout builds a working playbook from your website alone, and you can add documents later from the Playbook page._ Skip with the skip button in the stepper, or continue with the footer. Nothing is lost by coming back to this later. ## Adding documents later Open **Playbook** in the sidebar. The [Rules](/playbook/rules) tab is where extracted rules live, and where you add your own. Documents added after setup go through the same reading and produce rules the same way. ## What happens next Step 4 asks you to [confirm competitors](/setup/confirm-competitors), the list Scout built while reading your search market. --- # Confirm competitors URL: https://docs.audience.so/setup/confirm-competitors > Step 4 of setup and the Competitors settings page. Scout proposes who you are measured against; you confirm the set, because it decides share of voice, gaps and part of every score. Scout does not guess who you compete with. While reading your search market in step 1, it looked at who else ranks for the queries you already rank for, and step 4 shows you what it found. The step opens with **Who are you measured against?** and explains: _Scout found these by looking at who else ranks for the queries you already rank for. Nothing here needs changing unless something looks wrong — drop anyone who is not really a competitor, and add anyone Scout missed._ ![Step 4 of setup, the competitor list with classifications](/docs/screens/onboarding-competitors.png) _The six strongest matches first. Every edit saves itself._ ## Why the set matters The confirmed competitor set is used in three places, so a name that does not belong quietly distorts all three: - **Share of voice** in [AI visibility](/ai-visibility/where-you-stand) is a share of something, and this set is the something. - **Gaps.** Scout looks for searches where a confirmed competitor is strong and you are absent or weak. That is one of the signals behind the [opportunity queue](/opportunities/reading-the-queue). - **Part of every score.** The "what competitors already own" component of the [opportunity score](/opportunities/the-score) reads this set. Until competitors are confirmed, that component scores nothing and the product says _Scout does not know who you compete with yet._ ## Reviewing the list The six strongest matches appear first, with a **show all** control for the rest. Each row has a classification you can change: | Classification | Meaning | | ----------------------- | ---------------------------------------------------------------------------------------------------- | | **Direct competitor** | Sells something a buyer would weigh against you. | | **Search competitor** | Ranks for your topics without selling what you sell. | | **Reference publisher** | Scout cites these rather than trying to out-rank them: a publication, a directory, a standards body. | The definitions are written into the dropdown itself. Drop anyone who is not really a competitor, add anyone Scout missed by domain, and move on. **Your changes save as you make them.** There is no Save button. While discovery is still running the page says **Scout is still looking**. If it finished with nothing, **No competitors found yet**, and you can add them by hand. ## Editing later Open **Competitors** from the workspace menu. The settings page shows the whole list rather than only the strongest matches, because someone who navigated here usually came looking for one specific wrong name. ![The Competitors settings page](/docs/screens/settings-competitors.png) _The full list, with the same classifications and the same self-saving edits._ The page's own words: _Who Scout measures you against. This set decides your share of voice, which gaps Scout proposes, and part of every opportunity score — so a name that does not belong here quietly distorts all three._ Scout keeps proposing new competitors as your search market shifts, and proposals wait for you to confirm them. ## What happens next The footer reads **Next — build my workspace**. Step 5 is [the build](/setup/the-build). --- # The build URL: https://docs.audience.so/setup/the-build > Step 5 of setup. Scout puts the workspace together in a few minutes, task by task, and tells you what it learned when it is done. The last step is Scout's. It takes what the first four steps gave it and builds the workspace: the company profile, the playbook, the search baseline, the first queue of opportunities. The page opens with **Scout is putting your workspace together.** and a lede that tells you the one thing worth knowing: _This usually takes a few minutes. You can leave this page — nothing stops when you do, and Home will tell you when it is done._ ![Step 5 of setup, the build in progress](/docs/screens/onboarding-building.png) _Each task is a line with its own state, counters and note._ ## The tasks Seven tasks run, each shown as a line with a label, live counters where they apply, and a note. In the order they appear: | Task | What it does | | --------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------- | | **Reading your website** | Crawls your public pages and counts the pages read. | | **Analyzing your Search Console queries** | Pulls your search history when Search Console is connected; skipped otherwise. | | **Measuring your search market** | Takes the first ranking snapshot: the keywords you rank for today. | | **Working out what you sell and who buys it** | Writes the company profile, counting products and audiences identified. | | **Sizing up your competitors** | Analyses the competitor set you confirmed. | | **Writing down your editorial rules** | Builds the playbook from the standard rubric and any documents you added. | | **Finding what to move forward** | The first opportunity scan. It waits for the company profile, because a topic judged without one would be scored on search demand alone. | ## What each state means | State | Meaning | What to do | | -------------- | ------------------------------------------------------------------------------------------- | ---------------------------------------------------------- | | Working | Running now, with progress where Scout can estimate it | Nothing | | Done | Finished, with its counts | Nothing | | Skipped | You skipped the step it depends on (for example, Analytics) | Nothing, unless you want it: connect from Settings later | | Waiting on you | Scout needs a decision before it can continue, for example which Analytics property to read | Follow the link on the line, such as **Choose a property** | | Failed | The task stopped with an error, shown on the line | Press **Run it again** | A failure never takes the rest down. The other tasks finish, and the failed one waits for its retry. ## Leaving the page Everything here runs in the background. If you close the tab, Home shows a build panel with the current step, a progress bar, the counters that have landed, **Still coming: …** and **Waiting on you: …** with a link back into setup. The button on Home reads **Finish setting up**, and the page fills itself in as each piece lands. Anyone who is not an owner or admin sees the same progress with a note: _An owner or admin finishes setup from here._ ## When it finishes The heading changes to **Your workspace is ready.** and the page shows **Here is what Scout learned about {domain}.** with the counters laid out as a grid: pages read, queries pulled, competitors confirmed, rules written, opportunities found. One button: **Review what Scout found**. It takes you to the [opportunity queue](/opportunities/reading-the-queue). If a task is still going, the button reads **Go to Home anyway**, with the note _Home works while this finishes. It fills itself in as each piece lands._ ## What happens next - Read the [weekly briefing](/home/your-weekly-briefing) on Home. - Work down the [opportunity queue](/opportunities/reading-the-queue) and make your first decisions. - Check the [company profile](/playbook/your-company-profile) Scout wrote, and set a [direction](/playbook/direction) if the business is heading somewhere the website does not yet say. - [Invite your team](/setup/invite-your-team). --- # Invite your team URL: https://docs.audience.so/setup/invite-your-team > How owners and admins invite people to a workspace, which role to give them, and what the invited person sees. Accounts in Audience are created by invitation. Nobody signs up on their own: a workspace owner or admin invites a person by email, and the invitation is the door in. This keeps a product that spends money on your behalf behind people you chose. ![Workspace settings with the members list and pending invitations](/docs/screens/settings-workspace.png) _Members, their roles, and the invitations still waiting to be accepted._ ## Sending an invitation 1. Open the workspace menu in the sidebar (click the workspace name) and choose **Workspace**. 2. Under **Workspace members**, press **Invite member**. 3. Enter the person's email address and choose a role: **Admin**, **Editor** or **Viewer**. Owner is not assignable here. 4. Send. The invitation appears under **Pending invitations** until it is accepted, and you can cancel it from there. Only owners and admins can invite. Editors and viewers see the members list but not the button. ## Which role to give | Role | Give it to | What they can do | | ------ | ---------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------- | | Admin | A second person who runs the workspace | Everything except deleting the workspace: setup, integrations, the playbook and company profile, the AI visibility panel, members and roles | | Editor | The people who decide topics and review drafts | Approve and reject opportunities, research topics, approve outlines and drafts, comment, add topic exclusions | | Viewer | Stakeholders who need to see, not decide | Read everything, comment on drafts | Every member can read everything in the workspace. The full matrix is in [Who does what](/start-here/who-does-what). ## What the invited person sees The email carries a link. What happens next depends on whether they already have an account. - **A new person** is taken to **Create your account**. Their invitation is attached to the sign-up, so when they finish they land in your workspace. Without a valid invitation link, the registration page says: _Accounts are created by invitation. Ask a workspace owner to invite you, then open the link in that email._ - **Someone who already has an account** signs in and sees a **Pending workspace invitations** window on Home: _Accept or decline the workspaces you have been invited to join._ Each row names the workspace and who invited them, with **Accept** and **Decline**. Once accepted, the workspace appears in their switcher and they can move between workspaces from the sidebar. ## Managing members afterwards From the same **Workspace** page: - Change a member's role from the dropdown on their row. Owners' roles are fixed. - Remove a member with the remove control on their row. Owners cannot be removed. - Cancel an invitation that has not been accepted from **Pending invitations**. Expired invitations are deleted automatically. A person can leave a workspace they belong to from **Settings → Workspaces**, using the leave control on its row. Owners cannot leave; a personal workspace cannot be left either. ## Related - [Workspace and members](/settings/workspace-and-members) for workspace name, rebuild cadence and deleting a workspace. - [Account and security](/settings/account-and-security) for the invited person's own profile, password and two-factor settings. --- # Your weekly briefing URL: https://docs.audience.so/home/your-weekly-briefing > Home answers four questions in order. What should I do this week, what is waiting on me, is any of this working, and what has Scout figured out. Home is not a dashboard. It is the one page to read on a Monday. It answers four questions, in this order: what should I do this week, what is waiting on me, is any of this working, and what has Scout figured out. Everything on it links to the place where you act. ![Home for the demo workspace](/docs/screens/home.png) _The masthead counts what is worth moving forward. The button takes you to the queue._ ## The masthead The eyebrow reads **Week of {date}**. The heading counts: **Nothing needs you this week.**, **One thing worth moving forward this week.**, or **{n} things worth moving forward this week.** Below it, one primary button: **Review what Scout found**, with the line that describes how Audience works, _Approve the thinking before the writing._ It opens the [opportunity queue](/opportunities/reading-the-queue). ## While the workspace is still being built During setup the heading changes to **Scout is still getting to know you.** and the button to **Finish setting up**, with the note _A few minutes now, and this page fills itself in._ A build panel shows the current step, a progress bar, the counters that have landed, **Still coming: …**, and **Waiting on you: …** with a link back into setup. The panel disappears once [the build](/setup/the-build) is done. ## The four numbers Four cards decide whether the week is on track. Each one says what it cannot measure rather than showing a zero it did not earn. | Card | What it shows | When it cannot | | ------------------------ | ------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Open opportunities** | Found and waiting on a decision | _Nothing found yet_ | | **Awaiting your review** | Outlines and drafts that need a call | _Nothing waiting_ | | **Organic clicks** | The last 30 days, with a trend against the 30 before | _Connect Search Console_ | | **Business outcome** | A label taken from your Analytics property, for example _Signups from organic_ | _Choose a property_ if Analytics is connected but no property is chosen; otherwise _Measure outcomes, not output. Connect Google Analytics and Scout reports on signups or revenue instead of clicks._ | ## What to move forward The top five to seven recommendations, numbered, best first. Each shows its score, a type badge, a one-line reason, one measured evidence chip, and a **{n} sources** button that opens the sources behind it in a drawer so you can check the evidence without leaving the page. The badges say what kind of work each one is: | Badge | Meaning | | ------------- | ------------------------------------------------------- | | New page | A topic you have not written about | | Refresh | An existing page worth updating | | Slipping | A page losing ground it used to hold | | Nearly there | A search you almost rank well for | | Gap | A search a confirmed competitor wins and you do not | | Cluster | Several related searches that one article should answer | | AI visibility | A buyer question an assistant answered without you | **See all** opens the queue. Before the first scan the section says _Scout has not found anything yet_ and explains that the first scan runs once your site is crawled and Search Console is connected. ## What Scout learned The right rail carries one to three insights Scout drew from your own decisions and numbers, each with a headline, a line of detail and **Noticed {relative time}**. Empty at first: _Nothing learned yet. Scout picks up patterns from what you approve, edit and reject. Give it a few decisions and it will start telling you what it noticed._ Findings with evidence attached live under [Reports](/reports/overview). ## Running now While anything is running, live cards show each run's label, its current step and a progress bar. The page refreshes itself every few seconds until they settle, so you can leave it open and watch a scan land. ## What Scout did this week The last six finished runs, each with its outcome or, if it stopped early, its error and a **Retry** button. Retrying picks up where the run stopped rather than starting over. The full record of every run and every agent call is on the [Activity](/automation/activity) page, reached from the indicator at the foot of the sidebar. ## Invitations If you have been invited to another workspace, a **Pending workspace invitations** window opens when you arrive, with **Accept** and **Decline** for each. See [Invite your team](/setup/invite-your-team). --- # Reading the opportunity queue URL: https://docs.audience.so/opportunities/reading-the-queue > How the Opportunities screen is ordered, what each row tells you, and what the filters, badges and empty states mean. Opportunities is where Scout puts what it found. Every row is a topic it thinks is worth an article, with a score, the reason it thinks so, and the evidence behind that reason. The screen exists to be worked down from the top: read the thinking, then decide. ![The opportunity queue, best first](/docs/screens/opportunities.png) _The queue is ordered by score and nothing else._ ## How the list is ordered By score, highest first. There is no grouping and no pagination. The list does not reorder itself while you are reading it: a scan or re-judge that changes scores shows a progress banner at the top, and the list refreshes when the run lands. The heading counts what is open: "12 things worth moving forward." When the queue is empty it says "Nothing open right now." ## The lines under the heading Three short lines can appear under the heading, each telling you something about how the queue was built. **Judged against your company profile, version N.** Every topic was judged for fit against this version of the profile Scout holds for you. If the topics feel wrong for your business, the link beside it says **Wrong kind of topics? Tell Scout where you are heading**, which opens the [direction](/playbook/direction) fields. Once you have set a direction the line reads "and the direction you set" and links to **Review it**. **N topics kept out by your rules.** Topics a rule under [Topics Scout won't propose](/playbook/topics-scout-wont-propose) removed before scoring. The link takes you to the rules, so a rule that is too broad is one click from being found. **Scout set aside N topics that were not about your business before judging the rest.** A disclosure that expands to show the most-searched of them. See [Topics Scout set aside](/opportunities/topics-scout-set-aside). ## Filters The status chips carry counts, so you can see the shape of the queue before you click. | Chip | What it shows | | ---------------- | ----------------------------------------------------------------------------- | | **Open** | Topics waiting on a decision. This is the default view. | | **Flagged** | Topics that failed a guardrail. Kept out of the shortlist and out of the bin. | | **Shortlisted** | Topics you have marked as worth keeping in view without approving yet. | | **Approved** | Topics you moved forward. Each one is now an article in Content. | | **In progress** | Approved topics Scout is working on. | | **Not worth it** | Topics you rejected. | Below the status chips, the type chips narrow the list to one kind of signal. **All signals** shows everything; the others are the types described in [The score](/opportunities/the-score), such as a page that is nearly ranking, a page that is slipping, or a gap a competitor owns. The search box matches on the topic and the proposed title. ## What a row shows - **Type badge** and **status badge**. - **"{n} keywords, one article"** when Scout grouped several variations of the same question into one topic, so you commission one page rather than several that compete. - **The proposed title**, which is the row's link. - **The keyword** and its monthly search volume. - **The primary signal**: the strongest reason this topic is here, such as the position your page already holds for the query. - **Guardrail reasons**, inline, when the topic is flagged: "Fit with what you sell scored 28 out of 100, below the 40 this workspace treats as the minimum." - **The reason**, with clickable evidence. Clicking a piece of evidence opens the source drawer showing where the number or the claim came from. - **The score dial**, with the confidence range beneath it. ## Row actions Editors, admins and owners see actions on the right of each open row. None of them are optimistic: the row waits for the server to confirm before it changes. - **Move forward** approves the topic and starts research. See [Deciding](/opportunities/deciding). - **Not worth it** rejects it, with an optional reason. - **Do it anyway** appears on flagged rows instead. It overrides the guardrail and puts the topic in the queue. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). Viewers see the queue without actions. ## Research a topic The **Research a topic** button at the top right opens the research panel. It opens by itself when there is nothing in the queue, because then asking Scout to look at something is the only move on the page. See [Research a topic](/opportunities/research-a-topic). ## When the list is empty Three different things can be true when you see no rows, and the screen says which. **Everything Scout found is flagged.** Every topic failed a guardrail, usually fit with what you sell, so none reached the queue. The button **Review the flagged topics** switches to the Flagged filter, where each row says why it was held back. This is the state you will see if your company profile is thin or wrong, so also check [Your company profile](/playbook/your-company-profile). **Nothing matches those filters.** A filter or search term excluded everything. Clear it to see the whole queue. **No opportunities yet.** Scout has not found anything. The first scan runs after your site is read and, ideally, Search Console is connected. Confirming your competitors gives it more to work with. See [Set up your workspace](/setup/set-up-your-workspace). ## Related - [The score](/opportunities/the-score) - [Deciding](/opportunities/deciding) - [Your weekly briefing](/home/your-weekly-briefing) shows the top of this queue on Home. --- # The score URL: https://docs.audience.so/opportunities/the-score > What a score out of 100 is made of, how to read the eight components on an opportunity''s page, and why "not measured" is never a zero. Open any row in the queue and you reach the opportunity itself: the proposed title, the angle Scout suggests, why it recommends the topic, and the full working behind its score. Nothing on this page is a black box. Every point is attributed to a component, every component says what it measured, and anything Scout could not measure says so. ![One opportunity with its score breakdown](/docs/screens/opportunity-detail.png) _Each component shows the points it earned of the points available, and the evidence behind them._ ## The top of the page The proposed title is the heading. Under it sit the keyword, its monthly searches and the primary signal, then the **angle**: the approach Scout thinks an article on this topic should take, written for the searcher rather than for the keyword. If the topic failed a guardrail, a panel headed **This is below your guardrails** comes first; see [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). **Why Scout recommends it** gives the reasons in plain sentences. Each reason carries the evidence it rests on, and clicking the evidence opens the source drawer. ## How Scout arrived at the score The score is out of 100 and is built from eight components. Each one is shown with the points it earned of the points available, an explanation in words, and the evidence it used. | Component | What it asks | | --------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Fit with your business** | Is this topic about the market you sell into, would an article on it lead to something you sell, and is the person searching it someone you sell to? Judged against your [company profile](/playbook/your-company-profile). | | **What your own search data says** | What Search Console shows for this query: your position, impressions and clicks. | | **How many people search for it** | Monthly search volume across the grouped keywords. | | **Whether you can rank for it** | Keyword difficulty set against how strongly your domain ranks today. | | **What competitors already own** | Whether a confirmed competitor ranks for this and you do not. | | **Whether it collides with your pages** | Whether you already have a page that answers this, and how closely. | | **Where demand is heading** | Whether searches for this topic are rising or falling. | | **Outcome and AI visibility** | Whether this topic is one your buyers ask assistants about, and whether an answer named you. | Fit carries the most weight. A topic with huge demand and no fit is flagged, not shortlisted, because search volume alone knows nothing about your business. ### How each number is normalised Under every component is a line saying how the raw measurement became points, in words rather than formulas. Three you will see often: - **"Log-scaled, so head terms cannot dominate."** A keyword with ten times the searches does not earn ten times the points. - **"Judged by Scout, then bounded by deterministic checks."** Fit is Scout's judgement, but it is capped where the audience does not match: a topic can be no more relevant to your business than it is to your buyers. - **"Capped: Search Console is not connected."** Without your own search data the component cannot rise past a ceiling, and the page says so. ### Not measured is not zero When Scout could not measure a component, the component shows a **not measured** badge and the line "Not counted against the score. Its weight moved to what Scout could measure." The score is still out of 100; it simply rests on fewer measurements. This is why a score can be high with Search Console disconnected and why the same topic can change when you connect it. A topic scored before Scout began recording its working says: "This opportunity was scored before Scout started recording its working. It will explain itself after the next scan." ### The confidence range Beside the score is a range, such as 54 to 68. The narrower the range, the more of the score rests on measured components with high confidence. A wide range means Scout is reasoning from less, and the detail below says from what. ## Against what you already publish The sticky rail on the right holds the score and a second panel, **Against what you already publish**. This is the cannibalization check: whether writing this topic would compete with a page you already have. - A risk badge and a one-line summary. - **Closest existing page, N% similar.** - **Search Console overlap**, or "not connected" when Scout cannot read it. - **Your URLs ranking** for this query. - The related pages themselves, so you can open them. A high risk does not block a topic. It changes the recommendation: refresh the page you have rather than write a new one. ## One article, N searches When the topic groups several keywords, the page lists them under **One article, N searches**: "These are variations of the same question. Scout grouped them so you commission one page instead of competing with yourself across several." The number is the combined monthly demand. ## The action bar At the bottom, when the topic is open and you have permission: **Move forward**, **Find another angle** and **Not worth it**. See [Deciding](/opportunities/deciding). ## Related - [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics) - [Connect Search Console](/setup/connect-search-console), the single largest change you can make to how scores are measured. --- # Guardrails and flagged topics URL: https://docs.audience.so/opportunities/guardrails-and-flagged-topics > The four thresholds a topic must clear to reach the queue, what happens when it does not, and how to put a flagged topic through anyway. Scout never throws a topic away because a number was too low. A topic that fails a guardrail is **flagged**: kept, attributed to the rule it broke, and one click from the queue if you disagree. This page explains the rules, where flagged topics live, and how an override works. ## The four guardrails Each guardrail is a named threshold, evaluated on its own. A topic that fails one is flagged with that rule named, so you always know which line it crossed. | Guardrail | What it checks | | --------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Minimum search volume** | Monthly searches below which a topic is not worth an article. | | **Maximum keyword difficulty** | Difficulty above which your domain is unlikely to rank today. | | **Minimum opportunity score** | The composite score below which a topic is not shortlisted. | | **Minimum fit with the business** | How well the topic fits what you sell, as Scout judged it against your company profile. Without this, a topic with no connection to your business could reach the queue on search demand alone. | Two things follow from how they are applied. A guardrail only judges a measurement that exists. A topic with unknown search volume does not fail the volume guardrail; treating missing data as a failure would say something about your data, not about the topic. A guardrail flags and never drops. Every flagged topic is still on the Flagged filter with its score, its reasons and the exact numbers it was measured against. ## Where flagged topics live On the Opportunities screen, the **Flagged** chip carries a count. Rows there show the failed guardrail inline: "Fit with what you sell scored 28 out of 100, below the 40 this workspace treats as the minimum. Search demand alone carried this one." When every topic Scout found is flagged, the default view says so and offers **Review the flagged topics**. On the topic's own page, a panel headed **This is below your guardrails** lists each failed rule with the observed value beside the threshold, and the sentence explaining it. The panel reads: "Scout kept it out of the shortlist rather than throwing it away. If the threshold is wrong for this topic, put it in the queue anyway." ## Do it anyway The override lives in that panel and on the row: **Do it anyway**. There is an optional field, "Why is this one worth doing anyway?", and your reason is kept with the topic. Once overridden, the topic moves to the queue as an open topic and the page records who overrode it and when: "Overridden by Dana on 14 September." An override is a decision, and Scout treats it as one: a later rebuild records what the topic still fails but never sends it back to the flagged pile. Editors, admins and owners can override. Viewers see the panel without the button. ## Flagged topics can clear on their own Scores refresh daily as new search data arrives. A topic flagged for low search volume can read above the minimum next month; a topic flagged on fit can clear after you edit your company profile and Scout re-judges it. When a flagged topic clears every guardrail, it returns to the queue by itself. When an open topic stops clearing one, it is flagged again. Only topics nobody has decided on move this way. Shortlisted, approved and rejected topics stay where a person put them. ## When the fit guardrail flags everything If the Flagged filter is full and the queue is empty, the most common cause is the company profile, not the topics. Read the fit reasons on a few rows: if they say the searcher is a consumer or the topic shares only a word with your category, and that reading is right, the topics were never worth writing. If the reading is wrong, fix what Scout believes about you under [Your company profile](/playbook/your-company-profile) or set a [direction](/playbook/direction), and Scout will re-judge every open topic. ## Related - [The score](/opportunities/the-score) - [Direction and re-judging](/opportunities/direction-and-re-judging) --- # Research a topic URL: https://docs.audience.so/opportunities/research-a-topic > How to ask Scout to look into a topic, a search term, a keyword list or a competitor, and where the results land. The queue is what Scout found on its own. Research is what you ask it to find. Give it a topic, a term, a list or a competitor's domain, and a couple of minutes later anything worth writing is in the queue beside everything else, judged by the same measures. ![The research panel, open above the queue](/docs/screens/opportunity-research.png) _Research something: a topic, a search term, a competitor._ ## Opening the panel Click **Research a topic** at the top right of Opportunities. The panel is headed **Research something**: "Give Scout a topic, a search term, or a competitor. It will find what people search around it and add anything worth writing to this list." When the queue is empty the panel is already open, because asking Scout to look at something is the only move on the page. ## What you can give it - **A topic or a term.** Type it into the box, for example "deepfake detection api". Scout finds what people search around it. - **A list.** Paste one term per line. The counter beside the button shows how many terms will be researched; a request takes up to 25. - **A competitor.** Open **Compare against a competitor instead** and enter a domain. "Scout looks for what they rank for and you do not." Then click **Research this**. The panel says "Takes a couple of minutes. You can keep working." ## What happens next Scout expands what you gave it into the searches people actually make, groups variations into single articles, judges each group against your company profile, scores it and applies the guardrails. Anything that clears them appears in the queue as an open topic; anything that does not appears under Flagged, with the reason. A progress banner shows the run while it lasts, and the list refreshes when it lands. A researched topic is scored exactly like a discovered one, so the two sit side by side and can be compared honestly. ## Who can research Research spends provider and model budget, so it needs the same permission as approving an opportunity: editors, admins and owners. A viewer sees the panel with the note "Only editors and above can start research." ## The second way in The [AI visibility](/ai-visibility/questions-sources-and-market) Questions tab has its own research action. A question your buyers asked an assistant, which the assistant answered without naming you, can be sent to research on its own or in a batch of up to 25 from the Absent lane. It arrives in this queue as a topic, carrying the move the assistant's answer pointed at, and the question is marked "Sent to research" so it is not offered twice. ## Related - [Reading the queue](/opportunities/reading-the-queue) - [Deciding](/opportunities/deciding) --- # Deciding URL: https://docs.audience.so/opportunities/deciding > What each decision on an opportunity does, from Move forward to Not worth it, and what Scout learns from it. Every opportunity waits on one of three decisions. Scout does not make them. It makes its case, shows its evidence, and leaves the choice to a person, because the choice is the part that teaches it what you want. ## Move forward **Move forward** approves the topic. Scout starts researching it straight away: what searchers want, what the pages that rank already cover, what nothing covers, facts with a source. The topic becomes an article in [Content](/content/the-pipeline) at its first stage, and its status here changes to Approved, then In progress. Approving is the point at which Scout starts spending real effort on a topic, so it is available to editors, admins and owners. It is not undone from this screen; an article you no longer want is dealt with in Content. ## Not worth it **Not worth it** rejects the topic. A dialog asks "Why is this not worth it?" The reason is optional but useful: "it is the fastest way to teach Scout what you do not want. It reads every rejection." Scout weighs topics like a rejected one lower in future. The dialog also offers **Keep topics like this out from now on**. Ticking it writes a rule that removes matching open topics now and stops Scout proposing them again. You choose how the rule matches: | Match | What it removes | | ----------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- | | **exactly this keyword** | Only this phrase. | | **anything mentioning this** | Any topic whose keyword contains the phrase. | | **anything about this theme** | Any topic that is really about this subject, whatever words the search uses. "Scout reads each candidate and keeps out the ones really about this." | The rule appears under [Topics Scout won't propose](/playbook/topics-scout-wont-propose), where you can see how many topics it has removed and delete it. Rejected topics stay on the **Not worth it** filter. ## Do it anyway On a flagged topic the row action is **Do it anyway** instead. It overrides the guardrail and puts the topic into the queue as open, with your optional reason recorded. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). ## Find another angle On the topic's own page, **Find another angle** asks Scout to propose a different approach and title for the same topic without changing its score. Use it when the topic is right but the way in is not. ## Shortlisted **Shortlisted** is your holding state: a topic you want to keep in view without approving yet. It stays out of the open count, keeps its score up to date, and is never moved by Scout. Only a person moves a topic in or out of the shortlist. ## What a rebuild changes, and what it never touches Scout rebuilds the queue on the cadence set for your workspace, and refreshes scores on open topics every day. Two rules govern what that can move. Open topics move. A topic nobody has decided on can rise or fall in score, can be flagged if it stops clearing a guardrail, and can return to the queue if it clears one again. Decided topics do not. Approved, in-progress and rejected topics keep the score they were judged on, because a score that moved after you acted on it would explain nothing and rewrite the record of why you acted. A shortlisted topic is re-scored, since its score is still information you are weighing, but its status is left alone. ## Related - [Reading the queue](/opportunities/reading-the-queue) - [Topic and outline review](/content/topic-and-outline-review), the next screen after Move forward. --- # Direction and re-judging URL: https://docs.audience.so/opportunities/direction-and-re-judging > Why the queue changes when your company profile changes, and what Scout re-judges and leaves alone. Every topic in the queue was judged for fit against one thing: what Scout believes about your company. Change that, and the judgements are exactly what is now out of date. So a change to the company profile or to your direction sends every open topic back in front of the judge. ## Where the judgement comes from The line under the Opportunities heading says which version it was: "Judged against your company profile, version 3." The profile itself lives under [Your company profile](/playbook/your-company-profile). It is built from your website and any documents you gave Scout, and every open topic's fit score, and the reasons beside it, rest on that reading. If the queue is full of the wrong kind of topic, the profile is the first place to look. The link beside the version number says so: **Wrong kind of topics? Tell Scout where you are heading**. ## Direction outranks the website Your website says what you sold last year. [Direction](/playbook/direction) is where you say what the business is doing now, in two fields: **Focusing on** and **Moving away from**. Scout weighs a topic against your direction before anything it read on the site, so a topic that serves a product you are leaving reads as low fit even while the product page is still live. ## What a re-judge does Saving a new profile version or a new direction starts a re-judge. Scout puts every open topic back in front of the judge, records the fresh reading, and re-scores the queue. A banner shows the pass while it runs; the list refreshes as topics are judged. What can move: - **New topics** can be flagged if their fit falls below the minimum, and their scores and reasons change. - **Flagged topics** can return to the queue if their fit now clears the guardrail. - **Shortlisted topics** are re-scored, because their score is information you are still weighing, but they stay shortlisted. What stays put: - **Approved and in-progress topics** keep the score and the reasons they were approved on. - **Rejected topics** stay rejected. - **Overridden topics** stay in the queue. An override is a decision, and a re-judge records what a topic still fails without undoing it. When the pass finishes, the version number under the heading advances, and the completion note says what moved: how many topics went up, went down, were newly flagged or cleared a guardrail. ## After a re-judge that flags most of the queue Read the fit reasons on a few flagged rows before changing anything. Each one names the searcher ("Someone learning how something works, not shopping"), the fit with your market, and whether an article could lead to a product of yours. If those readings are right, the topics were never yours, and the fix is a better queue rather than a lower bar: research the topics you know matter, and let the next rebuild work from the corrected profile. If the readings are wrong, correct the profile field they rest on and Scout will re-judge again. ## Related - [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics) - [Topics Scout set aside](/opportunities/topics-scout-set-aside) --- # Topics Scout set aside URL: https://docs.audience.so/opportunities/topics-scout-set-aside > Why some candidate topics never reach the judge, how to see which ones, and what to do if one of them is yours. Before Scout judges a topic for fit, it asks a quicker question of every candidate it found: is this plainly not about your business? The ones that plainly are not, it sets aside. They are never scored, never flagged, and never take a place in the queue. This page explains why that step exists and how to check it. ## Why it exists The searches around a company's topics are wide. Expand "account sharing" and you get streaming device limits; expand "email verification" and you get webmail login pages; expand a product name and you get things that merely share a word with it. Those searches have huge volumes, and volume alone once filled a queue with two hundred topics that had nothing to do with the company that owned it. Every one was correctly flagged for poor fit, and the queue read as empty, while the topics that belonged, with a thousand searches a month, were never even looked at. Setting aside the plainly off-market candidates first means the queue is chosen from what could be yours, not from what is most searched. ## What gets set aside Scout asks, for each candidate, whether your company would have a reason to publish an article on it for the people it sells to. It sets a topic aside only when the answer is a confident no: a consumer's errand, another company's brand or product, a general fact, a word that only resembles your vocabulary. Anything uncertain goes on to the full judgement. If Scout cannot reach an answer for a batch, it keeps the whole batch rather than guess. ## Seeing what was set aside After a scan, a line appears under the Opportunities heading: **Scout set aside 2,670 topics that were not about your business before judging the rest.** Expand it to see the most-searched examples and a note: "If one of these belongs to you, research it by name and Scout will judge it properly." The examples are the highest-volume topics set aside, because those are the ones a person would most want to check. ## If one of them is yours Use [Research a topic](/opportunities/research-a-topic) and give Scout the topic by name. A researched topic goes straight to the full judgement, is scored like any other, and appears in the queue or under Flagged with its reasons. If Scout keeps setting aside topics that are genuinely yours, the cause is usually a company profile that does not mention that part of your business; see [Your company profile](/playbook/your-company-profile). ## Related - [Reading the queue](/opportunities/reading-the-queue) - [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics) --- # The content pipeline URL: https://docs.audience.so/content/the-pipeline > How the Content page groups every article by stage, shows who it is waiting on, and how each article's own record explains what happened and who decided it. Once you move an opportunity forward, it becomes an article in production. The **Content** page is where every one of those articles sits, from the moment Scout starts researching to the moment an approved version leaves as a download. It is built to be read from the top: the stages that need a person come first, then what Scout is working through, then what is finished. ![The content pipeline, grouped by stage](/docs/screens/content.png) _Stages that are waiting on a person come first. Scout's own work follows._ ## The header count The line under the heading says the one thing you need to know before you read anything else: - **"{n} articles are waiting on you."** when any stage needs a decision from a person. - **"{n} articles in production. Nothing is waiting on you."** when Scout has everything it needs. - **"Nothing in production yet."** before the first opportunity has been approved. ## Finding an article The search box looks across title, keyword, topic and owner. The stage chips underneath show **All** plus every stage that currently has something in it, each with a count. Stages that are waiting on a person are highlighted so they are the first chips you see. Filtering to a stage or a search term that matches nothing says so: _"Nothing matches. Clear the search or pick another stage."_ ## Reading a stage Each stage is a section with its heading and count. A stage that needs a person carries a **Waiting on you** badge. Sections show five rows and then a **Show all {n}** control, so a long backlog does not push the next stage off the screen. A row shows the title, the keyword and its monthly searches, a **Refresh** badge when the article updates an existing page rather than adding a new one, the assignee, and **Score {n}**, the opportunity score the article was approved on. Three kinds of row look different on purpose: - **Rows waiting on a person** show how long they have waited, as **waiting 4h** or **waiting 3d**, and the next action: **Review the topic**, **Review the outline** or **Review the draft**. The wait turns amber after 48 hours. - **Rows Scout is working on** show the current step and a progress bar rather than a spinner, so you can tell "writing the outline" from "checking sources". - **Failed rows** are not links. They show what went wrong, or _"Scout could not finish this step."_ when there is nothing more specific to say, and a **Try again** button that picks up where the step stopped. ## The stages | Stage | Who it waits on | What is happening | | ----------------- | --------------- | --------------------------------------------------------------------- | | Researching | Scout | Reading what ranks, what searchers want and which facts have a source | | Topic review | You | Confirming the topic and angle before anything is written | | Writing outline | Scout | Turning the research into a structure | | Outline review | You | Approving the thinking before the writing | | Drafting | Scout | Writing the article from the approved outline | | Draft review | You | Reading the draft, its score and its flags, then deciding | | Changes requested | Scout | Revising against the notes and flags you sent back | | Approved | You | Ready to leave as Markdown | | Publishing | Scout | Handing the approved version to a destination | | Published | Nobody | Live, and now tracked in Reports | | Failed | You | A step stopped; **Try again** restarts it | Three of these wait on a person: Topic review, Outline review and Draft review. They are the ones the **Waiting on you** badge marks. ## The empty state Before the first approval the page says **"Nothing in production yet"** and tells you what fills it: _"Approve an opportunity and Scout starts researching it. Everything it produces from that point, research, outline, draft, appears here at the stage it has reached."_ The way in is the [opportunity queue](/opportunities/reading-the-queue). ## An article's record Open any row and you reach the article's own page. Its job is stated in one line: what happened to this, and who decided it. ![One article's record](/docs/screens/content-item.png) _Every version, review and decision, in one place._ At the top: the status badge, a **Refresh** badge where it applies, the quality score badge once a review has run, the title, the keyword with its monthly searches and audience, and the **assignment** control. Owners, admins and editors can set an owner and a due date here. Below sits the goal the article was approved to meet. The buttons go to the work: **Open the review**, **Outline**, and **Published page** once the article is live. Then the sections: - **What this was built from.** The search intent, the recommended angle, a count of sources, and **See what Scout read**, which opens a drawer titled "Research behind this article" showing every source Scout consulted before it wrote anything. - **Reviews.** Every review that has run, with its score, its routing label, the rubric result as x/y, and how many flags are still open. - **The feedback thread.** Read-only here; you write into it from the review screen. - **Drafts and Outlines.** Every version, labelled **Written by Scout** or with the editor's name, with an **Approved** badge on the one that counts and a summary of what changed. - **Decisions.** Who approved or sent back what, and when. - **Publication.** Pending, Draft created, Published or Failed. While Scout is the one moving the article, the page refreshes itself every few seconds, so you can leave it open and watch a step land. ## What happens next An article that has just entered production goes to [topic review, then outline review](/content/topic-and-outline-review). One with a draft goes to [draft review](/content/draft-review). One that is approved goes to [publishing](/content/publishing-and-markdown). --- # Topic and outline review URL: https://docs.audience.so/content/topic-and-outline-review > The two checkpoints before a word of the article is written, where you confirm the topic Scout researched and approve or redirect the outline it drafts from. Audience asks you to approve the thinking before the writing. That is two checkpoints, and both come before the draft exists: a short one where you confirm what Scout found in research, and a longer one where you read, edit and approve the outline. Everything Scout writes afterwards is written from the outline you approved. ## Topic review When research finishes, the article moves to **Topic review** and the row on the [Content page](/content/the-pipeline) says **Review the topic**. What you are confirming is the shape of the piece now that Scout has read what ranks: the search intent it found, the angle it recommends, and the sources it is working from. If the research changed your mind about the topic, this is the cheapest moment to stop. Confirming sends Scout on to write the outline. ## Outline review ![Outline review](/docs/screens/outline-review.png) _The outline on the left is editable. What you save becomes the version Scout drafts from._ The page opens with the instruction it is built around: _"Approve the thinking before the writing. Edit anything here, what you save becomes the version Scout drafts from."_ ### What Scout checked Before you read the outline, Scout reports its own pre-checks, each marked as a pass or a failure: whether the outline answers the search intent, covers what nothing ranking covers, avoids the topics your playbook rules out, and so on. A failure is information, not a lock. The panel says so: _"You can approve anyway, these are Scout's observations, not a lock."_ Where the outline is close to something you already publish, a duplication-risk warning appears here too, so a page that would compete with your own is caught before it is written. ### The outline The outline is Markdown in an editor. It is editable when you can approve the article and it is at the outline review stage; otherwise it is read-only. Change a heading, drop a section, reorder the argument. The version you save is the version Scout drafts from, so an edit here is the most direct way to steer the draft. ### The research rail Beside the outline sits what the outline was built from, so you can check the thinking against its evidence: - **What searchers want.** The intent behind the keyword, in a sentence. - **Recommended angle.** The way in that Scout proposes, and why. - **What nothing ranking covers.** The gap the article can own. - **Facts with a source.** Each claim Scout intends to use, with the source it came from. A claim that needs care is marked _"Handle with care: …"_ with the reason. - **Questions readers ask.** The follow-up questions the piece should answer. - A count of the sources Scout read. ### Versions Every outline is kept. The **Versions** rail lists each one, labelled **Written by Scout** or with the editor's name. Nothing is overwritten; see [versions and approval](/content/versions-and-approval). ### Deciding The decision bar stays at the bottom of the screen. It has a feedback box labelled _"Ask for a different outline, optional, the current one is kept either way"_, and two buttons: - **Approve outline & start drafting.** Scout begins the draft from the outline as it stands, including your edits. - **Write it again.** Scout writes a new outline. Put what you want changed in the feedback box first, for example _"Lead with the compliance risk, and drop the tooling comparison."_ The current outline stays in the Versions rail. While Scout is writing, the page says _"Scout is writing the outline now."_ and updates itself when the outline is ready. ## Who can do this Owners, admins and editors approve outlines and ask for another. Viewers can read the outline and its research. See [who does what](/start-here/who-does-what). ## What happens next Approving the outline starts the draft. When it is ready, the article moves to [draft review](/content/draft-review). --- # Draft review URL: https://docs.audience.so/content/draft-review > The screen where you read a draft beside everything Scout found wrong with it, then make one of two decisions: send it back with notes, or approve it and move it forward. Draft review is the screen this product lives on. The article is on the left, Scout's findings are on the right, and two decisions sit at the bottom. Scout has read the draft against your playbook and a generic editorial rubric and said what it found and why. It has not changed anything on its own. The header says exactly that: _"Read it, then decide."_ ![Draft review with the findings rail](/docs/screens/draft-review.png) _Article left, findings right, your decision at the bottom._ ## The header The status badge, the version line (**Version 3 · Written by Scout**, or the editor's name when a person saved the last version), the title, and the keyword with its monthly searches and audience. Once a version has been approved, a **Download Markdown** button appears here with the line _"Approved and ready. Publishing is manual, put it live yourself."_ See [publishing and Markdown](/content/publishing-and-markdown). ## The article The draft is Markdown in an editor, with three fields above it for the **SEO title**, **meta description** and **slug**. You can edit any of it; a saved edit becomes a new version under your name rather than overwriting Scout's. Citation markers in the text open the source drawer at the cited source, so a claim can be checked without leaving the page. Selecting a passage, or a whole block, offers **comment on this**. Both are covered in [flags, comments and sources](/content/flags-comments-and-sources). ## Scout could not answer these Where research left a question Scout could not settle, it is listed here rather than guessed at in the draft: the question, and why it matters to the piece. These are the places a person's knowledge is needed before the article goes out. ## The review sidebar The right rail is Scout's reading of the draft. **Scout quality score** is out of 100. Under it, a routing line says what the score means for you: | Routing | What it says | | -------------------- | --------------------------------------------------------------------------------------------------------- | | **Ready for review** | _"Nothing is blocking this. Read it, then move it forward."_ | | **Needs revision** | _"Worth another pass before this goes out."_ | | **Blocked** | Names the flag that blocks it, for example _"16 of 28 claims could not be traced to a source Scout read"_ | Three smaller numbers explain the score: **Generic rubric x/y**, the standard editorial checks the draft passed; **Your playbook n%**, how much of your own rules it satisfied; and **Blockers n**. The source-status panel underneath shows how many claims trace back to a source Scout actually read. **What Scout found** is the list of flags. The panel tells you how to use it: _"Open one to see what to do about it. Ticked findings go back with the draft when you send it."_ A draft with nothing flagged says _"Nothing flagged. Read it and decide."_ A version Scout has not checked yet says so too. How to act on a flag is on the [flags page](/content/flags-comments-and-sources). ## Your call The decision panel stays in view as you read. It has a notes box, **What needs to change, if anything**, with a placeholder that shows the tone to use: _"Name the readiness work separately in the pricing section."_ Under it, a helper line counts what will travel with your notes: _"Tick flags in the right rail to send them back too."_ or _"{n} flags will be sent back with your notes."_ Two buttons: - **Send back.** The draft returns to Scout with your notes and the flags you ticked. The article moves to Changes requested, and Scout writes a new version against them. - **Approve & move forward.** The version you are reading becomes the approved version. A blocked draft cannot be approved until its blocker is applied or dismissed. On a phone the panel folds to one row until you tap **Send back**. ## After a send-back When Scout returns a revision, the feedback thread shows a revision summary: what it addressed, and what it did not, with a reason for each. A note it could not act on is said plainly rather than quietly skipped. Read the summary before the new version; it tells you where to look. ## Failed and missing drafts If a step stopped, the page shows the error and a **Try that step again** button. If there is no draft yet, it says so and offers **Write the draft**; while Scout is writing, it says _"Scout is writing the draft now."_ and refreshes itself. ## Who can do this Owners, admins and editors approve, send back, apply and dismiss flags, and vouch for sources. Anyone who can view the article can comment on it. See [who does what](/start-here/who-does-what). --- # Flags, comments and sources URL: https://docs.audience.so/content/flags-comments-and-sources > How to act on what Scout flagged in a draft, how to leave comments for your team, and how to check and vouch for the sources behind every claim. Three things sit beside a draft while you review it: the flags Scout raised, the comments your team left, and the sources every claim was built from. Each is decided on its own, and each decision teaches Scout something. ## Flags A flag is one finding from the review: a playbook rule the draft breaks, a rubric check it fails, a claim that cannot be traced. Each carries a severity. A **blocker** stops approval and opens itself when you arrive, so the thing standing in the way is the first thing you see. Open a flag to read what Scout found and what to do about it. Then choose one of three actions: - **Apply.** Scout's suggested change is made to the draft, as a new version. Nothing is applied without you choosing it. - **Dismiss.** The flag is wrong for this draft. A reason is required, and it is what Scout learns from: the dialog is titled **"Why is this not an issue?"** and asks you to _"Say what makes it wrong here rather than that it is wrong."_ The placeholder shows the shape: _"The spelling is deliberate, we write for a US audience."_ Once dismissed, the flag reads **Dismissed, Scout will remember**, and Scout stops raising it on future drafts. - **Tick it to send back.** A ticked flag travels with your notes when you press **Send back**, and Scout revises against it. The decision panel counts the ticked flags for you. Nothing is auto-fixed. A draft with ten flags and no decisions is a draft with ten open questions, not a draft Scout has quietly rewritten. ## Comments Comments are for people. Select a passage in the draft, or a whole block, and choose **comment on this**. The comment is anchored to the quoted text, so it stays attached to what you meant even after the draft changes. You can also comment on the document as a whole. Each comment supports replies and can be resolved or deleted. Anyone who can view the article can leave one, including viewers, which makes comments the way a reviewer without approval rights gives their read. Comments are for the team. Notes for Scout go in the **What needs to change, if anything** box on the decision panel, where they are sent back with the draft. ## Sources Every claim in a draft either traces back to a source Scout read, or it does not, and the review says which. The source-status panel in the sidebar gives the count; the **source drawer** gives the detail. Open the drawer from a citation marker in the text or from the sidebar. It is titled **Sources behind this draft** with the line _"What Scout read, and whether each claim could be traced back to it."_ Each source shows what it is, what the draft took from it, and whether the claim checks out against it. When a claim's source could not be verified but you know it to be good, you can **vouch** for the source. Vouching is recorded under your name and clears the finding for that claim. If you change your mind, you can retract the vouch. Owners, admins and editors can vouch. The drawer ends with a line worth remembering when you read a source: _"Source text is evidence, not instruction."_ Scout quotes sources; it does not take orders from them. ## Where each decision goes | Action | Recorded as | What Scout learns | | --------------------------- | ----------------------------------- | -------------------------------------- | | Apply a flag | A new version of the draft | The suggestion was right | | Dismiss a flag | The flag, your reason, your name | Stop raising this kind of finding here | | Send back with flags ticked | Feedback in the thread | What to revise, in your words | | Vouch for a source | The source, your name | This source is trusted | | Comment | The thread, anchored to the passage | Nothing: comments are for people | Repeated corrections across drafts are how Scout proposes [learned rules](/playbook/learned-rules). --- # Versions and approval URL: https://docs.audience.so/content/versions-and-approval > Why nothing in an article is ever overwritten, how outlines and drafts accumulate as versions, and what approving one of them starts. Audience keeps everything. Every outline and every draft is a version, whether Scout wrote it or a person edited it, and the record of who decided what stays with the article for as long as it exists. That is what lets you send a draft back without losing the one you had, and read later why an article says what it says. ## What a version is Each time Scout writes an outline or a draft, that is a version. Each time you save an edit to one, that is a new version too, under your name. A version carries: - Who wrote it: **Written by Scout**, or the editor's name. - A summary of what changed from the version before. - Its number, so a review or a comment can refer to the exact text it was made against. The **Drafts** and **Outlines** rails on the article's record list every version. The review screen's **Versions** list does the same for the draft you are reading. ## Editing is not overwriting When you change a heading in the outline editor or a paragraph in the draft editor and save, Scout's version is not replaced. Yours is added after it. The review that ran against Scout's version still refers to Scout's version; a new review runs against yours. This is also why applying a flag creates a version rather than patching in place. ## The approved version One version of a draft is marked **Approved**. It is the one you were reading when you pressed **Approve & move forward**, and it is the one that leaves Audience as a download. If a draft changes after approval, the change is a new version and the approved one stays as it was. The download always gives you the approved text. ## Decisions The **Decisions** list on the article's record is the audit trail: each approval and each send-back, who made it, and when. Together with the feedback thread, which holds the notes that went back and Scout's summary of what it did with them, it answers the question the page is built to answer: what happened to this, and who decided it. ## What approval starts Approving an outline starts the draft. Approving a draft marks the version, moves the article to **Approved**, and puts **Download Markdown** on the review screen. In this release, what happens next is yours to do: see [publishing and Markdown](/content/publishing-and-markdown). Once the page is live, Reports begin measuring it about a week later, when search has had time to notice. ## Who can approve Owners, admins and editors. A viewer can read every version and comment on it, but cannot approve. See [who does what](/start-here/who-does-what). --- # Publishing and Markdown URL: https://docs.audience.so/content/publishing-and-markdown > How an approved article leaves Audience as a Markdown file, how to put it live in your own CMS, and what Reports start tracking once it is. Publishing is manual in this release. Scout writes, checks and revises; a person approves; then the approved article leaves Audience as a file and you put it live yourself. Nothing Scout produces reaches the public without your hands on it. ## Download Markdown Once a version is approved, the draft review screen shows a **Download Markdown** button under the header, with the line _"Approved and ready. Publishing is manual, put it live yourself."_ The button is always available for an approved article, whatever else is configured. The file contains: - A short front matter block with the **title**, the **slug** and the **meta description** from the review screen's fields, so a CMS or a static site generator can read them without retyping. - The **body** as Markdown, with headings, lists, links and citations as they appear in the editor. It is the approved version, exactly. If someone edits the draft after approval, that edit is a new version and the download is unchanged until a new version is approved. See [versions and approval](/content/versions-and-approval). ## Putting it live A short checklist that fits most CMSs: 1. Create the page or post in your CMS and paste the Markdown body. Most editors accept Markdown directly or convert it on paste. 2. Set the URL from the **slug**, the page title from the **SEO title**, and the description from the **meta description**. 3. Check the links and the source attributions survived the paste. 4. Publish, and confirm the page is reachable at the address you expect. ## What Reports track afterwards Once the page is live at a URL on your site, Reports begin measuring it. Give it about a week: search takes that long to notice a new page and start sending it impressions. From then on the page appears under **Pages Scout published** in [Content performance](/reports/content-performance), with its clicks, impressions, position, and a verdict on what to do next. Its article record shows **Published** under Publication and links to the **Published page**. ## The Publishing settings page Under the workspace menu, **Publishing** describes where finished articles would go. In this release it says that publishing straight to your CMS is off, and offers no destination to add. The Markdown download is the way out, and it does not depend on that page. Details are on [publishing destinations](/settings/publishing-destinations). ## Who can do this Anyone who can approve a draft can download it: owners, admins and editors. See [who does what](/start-here/who-does-what). --- # What AI visibility measures URL: https://docs.audience.so/ai-visibility/what-ai-visibility-measures > How Scout asks the questions your buyers ask of AI assistants, reads whether you came up, and what each number on the AI visibility screens can and cannot tell you. Your buyers no longer only search. They ask. They put a question to ChatGPT, Gemini, Claude or Perplexity, read one answer, and act on it. If that answer names a competitor and not you, you were never in the running, and no search ranking would have told you. AI visibility is how Audience measures that. Scout asks the same questions your buyers ask, reads each answer, and records whether you came up, where in the answer you appeared, how you were described, and which pages the assistant built the answer from. It is the only section of Audience that you configure and spend against, which is why it sits on its own rather than inside Reports. ![The AI visibility overview: where you stand and this week's moves](/docs/screens/ai-visibility-overview.png) _The Overview tab answers one question first: do assistants mention you?_ ## What Scout reads in an answer For every question on your panel, every assistant, every run, Scout records four things. | What | Meaning | | ----------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------- | | **Named** or **Absent** | Whether your company appeared in the answer at all. A question that has never been run shows **Not asked**, which is different from losing it. | | Position | Where you sat inside the answer that named you. First, second, third. This is a place in a paragraph, not a search rank. | | Tone | How you were described, from −100 to 100. Blank when the run did not assess tone. | | Built from | The pages the assistant retrieved while answering, and the other companies it named instead of you. | The Overview turns these into a scorecard, [Where you stand](/ai-visibility/where-you-stand), and into the three most valuable fixes, [This week's moves](/ai-visibility/this-weeks-moves). The Questions tab shows every question and every answer. ## Two lanes AI visibility has two lanes, and they answer different questions. **Your panel.** The questions you confirmed, asked of the assistants on a schedule, read for your own company. This is the lane that measures you. It is built on the [prompt panel](/ai-visibility/build-your-prompt-panel), and everything in Overview, Questions and Sources comes from it. **The wider market.** A modelled estimate of how often your brand and your confirmed competitors are mentioned across a broad corpus of AI answers, with an estimate of how many people ask about your topics. It is broad and cheap, and it is not about your own questions. It lives on the Market tab, and it is measured only when you ask. ## What the numbers cannot tell you Answer engines do not repeat themselves exactly. Audience says so wherever it matters, and the caveats travel with the data rather than sitting in a footnote. - **One answer is one sample.** Each result is a single answer from one assistant on one day. Read a pattern across runs, never a single line. This is why the panel runs every morning rather than once a week. - **A blank tone is not a neutral one.** When the run did not score tone, the field is empty. Audience never renders an unassessed brand as zero, because a brand nobody assessed is not a brand assessed and found unremarkable. - **Position is within the answer.** "Average position 1.4" means you were usually named first or second in the answers that named you. It says nothing about search results. - **Market volume is modelled.** "AI search volume" on the Market tab is an estimate from a third-party corpus, not a count of real questions. One large brand in your confirmed competitor set can dominate that share. - **Coverage varies by assistant.** Figures for one of the assistants cover the US in English only, and the Market tab says so when it applies to you. ## When it is switched off AI visibility can be switched off for a workspace. The section still appears in the navigation, but the page says **AI visibility is switched off.** and explains that nothing is being measured or billed. Every action on the section is unavailable until it is switched back on. ## Related - [Build your prompt panel](/ai-visibility/build-your-prompt-panel) to choose the questions Scout asks. - [Costs and cadence](/ai-visibility/costs-and-cadence) for what runs on its own and what you pay for. - [Questions, sources and market](/ai-visibility/questions-sources-and-market) for the three tabs behind the overview. --- # Build your prompt panel URL: https://docs.audience.so/ai-visibility/build-your-prompt-panel > How to choose the buyer questions Scout asks the assistants every morning, keep the right suggestions, and confirm a panel that measures you fairly. The prompt panel is the set of questions you are measured on. Scout asks each confirmed question of an AI assistant every morning and reads the answer for whether your company came up, where, and how it was described. Because the panel decides what "visible" means for you, it is the one screen in Audience where a person commits the workspace to recurring spend, and nothing on it runs until you say so. Open it from the AI visibility section. The page is headed **Which questions should Scout ask?** ![The prompt panel, with the Your panel, Suggested and Dropped lanes](/docs/screens/ai-visibility-panel.png) _Nothing is ticked by default and nothing runs on arrival. Every proposal shows its reasoning._ ## Choose questions your buyers ask A good panel is made of the questions a buyer types before they know you exist: comparisons, requests for a recommendation, questions about the category, a problem they are trying to solve. For Resemble AI (demo) that is "What is the best voice cloning API for a startup?" or "How accurate are deepfake detection tools?" Never add a question that names you. "Resemble AI pricing" guarantees its own answer and tells you nothing about the buyers who have not heard of you. The page says the same thing in its introduction, and it is the most important rule on this screen. ## Suggest questions, or write your own **Suggest questions** drafts a panel from your company profile, products and audiences. It costs a model call, so it is a button and never runs automatically. Each suggestion arrives with its reasoning, so you can see why Scout thinks a buyer would ask it before you keep it. Before your workspace has finished its first build, the page says **Scout does not know enough about you yet.** A panel is derived from your profile, and a model asked to invent buyer questions for a company it has not read would write a convincing panel for a company that does not exist. Finish [the build](/setup/the-build) first. **Add a question of your own** is the box above the lanes. Type the question as a buyer would say it, and press **Add**. It goes straight into your panel. ## The three lanes The questions sit in three tabs. The page opens on whichever lane is asking for something. | Lane | What it means | | -------------- | --------------------------------------------------------------------------------------- | | **Your panel** | Asked on every run. Dropping a question keeps everything it has already measured. | | **Suggested** | Proposed by Scout, not yet kept. Nothing here is asked, or paid for, until you keep it. | | **Dropped** | Not asked, and not re-proposed. Restore one to put it back on the panel. | Each question shows an intent badge: **Comparison**, **Recommendation**, **Category**, **Problem** or **Pricing**. Next to it are the source (suggested or written by hand) and, once a question has been asked, when it was last asked. Expand a row to read Scout's rationale and, after a run, the domains the answer was built on. Once a question has been measured, its row also shows **You** (Named, Absent or Not asked), **Pos.**, **Tone**, **Named instead** and **Asked**, the number of times it has been run. ## Keep, drop, restore, confirm - In **Suggested**, **Keep** moves a question to your panel; drop sends it to Dropped, and Scout will not propose it again. - In **Your panel**, drop removes a question from future runs. Its history stays. - In **Dropped**, **Restore** puts a question back. Nothing takes effect until you press **Confirm panel** in the footer. The footer tells you what you are about to commit to: _"12 questions will be asked on every run."_ A panel can track up to 50 questions. Over that, the footer says **A panel can track 50 questions. Untick n to save.** and the button waits until you do. ## The budget line and Run now Above the list, a line shows how much of this workspace's monthly AI budget is spent, against its ceiling. Near the ceiling it turns amber and adds that the panel stops asking once the ceiling is reached, and reports the shorter sample it took rather than counting the unasked questions as misses. [Costs and cadence](/ai-visibility/costs-and-cadence) explains the rest. **Run now** asks every confirmed question today instead of waiting for tomorrow morning. It is an explicit spend, which is why it is a button. The first results usually land within a few minutes, and the Overview fills in as they do. ## What happens next Once a panel is confirmed, Scout asks it every morning. Read the results on [Where you stand](/ai-visibility/where-you-stand), and act on the losses in [This week's moves](/ai-visibility/this-weeks-moves). --- # Where you stand URL: https://docs.audience.so/ai-visibility/where-you-stand > How to read the AI visibility scorecard, from the share of questions you appear in to average tone and position, and what changes once two weeks of readings exist. **Where you stand** is the scorecard at the top of the AI visibility Overview. It condenses every answer on your panel into a handful of numbers, and each one carries a note about what it can honestly claim. ![The scorecard: questions you appear in, share of voice, tone, position and runs](/docs/screens/ai-visibility-overview.png) _Each assistant gets its own tile, so a bad week on one does not hide inside an average._ ## The numbers | Metric | What it says | | --------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Questions you appear in** | "1 of 50", with the percentage underneath. How many of your panel's questions the latest answers named you in. | | **Panel share of voice** | Across all the answers to your questions, the share of mentions that were you rather than the other companies named. Counted only across your own questions, so a brand outside your category cannot crowd the denominator. | | **Average tone** | How you were described, from −100 to 100. Blank when the run did not score tone. An unassessed brand is not a neutral one, and Audience never fills the gap with a zero. | | **Average position** | Where you sat inside the answers that named you. **Not placed** when no answer named you. A place in the answer, not a search rank. | | **Runs** | How many times the panel has been asked, and by which assistant most recently. Each run is one sample of a system that does not repeat itself exactly, not a census. | ## One tile per assistant Above the row of numbers, each assistant Scout asked has its own tile: the share of questions it named you in, and how many. For Resemble AI (demo) that might read Perplexity 2%, Gemini 6%, ChatGPT 10%, and it is normal for them to disagree. The assistants read different pages and reason differently, and the tiles are there so you can see which one is the problem. A tile lets you **ask one assistant again** without re-running the others, when one reading looks like an outlier and you want a second sample before you act. It is a spend, so it is a button. If an assistant refused a run, its tile keeps the last good reading and says the latest attempt failed, rather than showing zero. ## The trend Under the scorecard sits a line. For the first fortnight it is a per-run sparkline: every run is a point, and the caption says how many runs there have been and how far the share has moved since the first. Once two distinct weeks of readings exist, the line becomes a weekly trend, pooling each week's answers per assistant so a single odd morning cannot swing it. Read the trend, not the latest point. That is the whole reason the panel runs every morning rather than once a week: a single reading carries real variance, and four readings a month cannot separate a genuine change in visibility from the model phrasing itself differently. ## What to do with it The scorecard tells you how you are doing. The section beneath it, [This week's moves](/ai-visibility/this-weeks-moves), tells you what to do. For the full list of questions, who was named instead of you, and what the answers were built from, use the [Questions, Sources and Market tabs](/ai-visibility/questions-sources-and-market). --- # This week's moves URL: https://docs.audience.so/ai-visibility/this-weeks-moves > The three highest-value AI visibility losses, grouped by the kind of fix each one needs, and the actions Scout offers on each. A scorecard that says "2%" and stops has told you how you are doing and nothing about what to do, which is the half you came for. **This week's moves** is the other half. It sits under the scorecard on the Overview and picks the losses with the clearest fix, read off what the assistants actually said. The rest are under the Questions tab. ## How moves are chosen Scout looks at every question an assistant answered without naming you and reads the evidence in the answer itself: the pages it cited, the companies it named. From that it decides which kind of fix the loss points at, and ranks them. A page you already own that was cited but silent beats a page you would have to write, so those come first. At most three moves are shown, and a question you have already sent to research, or that has become an opportunity, is not shown again. ## The five kinds of move Moves are grouped by kind, and each kind is explained once above its questions. | Move | What it means | | -------------------------- | ---------------------------------------------------------------------------------------------------------------------------------- | | **Claim your citation** | The assistant built the answer from your own page and still never named you. Make the brand explicit on that page. | | **Improve your placement** | A roundup the assistant relied on already lists you, and it named others instead. The placement on that page is what it read past. | | **Comparison page** | The buyer was reading a comparison, and you were not in it. The comparison page is what the assistant would have to cite. | | **Earn a citation** | The answer leaned on a roundup somebody else wrote. Getting listed there moves the answer without outranking it. | | **Answer the question** | Nothing cited is a page you could join and no rival was named. Answer the question on your own site. | ## Reading a row Each row leads with the question itself, linked to its full answer on the Questions tab. Under it is one line of evidence that makes the row different from its neighbours: the assistant that answered, then what it found. - _Perplexity · Built from resemble.ai/blog/voice-consent_: your page, uncredited. - _ChatGPT · Named ElevenLabs, Play.ht instead of you_: a comparison you were left out of. - _Gemini · Built from a roundup on g2.com_: a page to earn a place on. - _ChatGPT · You are listed on techradar.com, and still skipped_: a placement to improve. Beside each row are the actions. **Open your page** or **Open {domain}** takes you to the page the assistant read. The research button carries the loss into your opportunity queue with the move attached as the recommended page: **Research a fix** for a citation or placement problem, **Research this comparison** for a comparison, **Research this** for the rest. What happens after that is described in [Research a topic](/opportunities/research-a-topic). ## Where the rest are Only three moves are shown here. Every other question you lost, with the same evidence and the same actions, is one tab away under Questions, where the **Absent** lane lists them all and lets you send up to 25 to research at once. See [Questions, sources and market](/ai-visibility/questions-sources-and-market). --- # Questions, sources and market URL: https://docs.audience.so/ai-visibility/questions-sources-and-market > The three AI visibility tabs behind the overview, from every question and answer to the pages assistants trusted and the modelled view of the wider market. The Overview is the summary. These three tabs are the evidence behind it: every question and what each assistant said, the pages the answers were built from, and a modelled view of the wider market that is deliberately kept apart from your own numbers. ## Questions **The questions you win and lose.** Every question on the panel, what each assistant said, and the move each loss points at. ![The Questions tab, with one question expanded](/docs/screens/ai-visibility-questions.png) _A row is one line until you open it. Then it shows the whole answer._ Above the list, a standing caveat says that each result is a single answer from one assistant on one day, and that answer engines do not repeat themselves exactly. Read a pattern across runs rather than any one line. **Lanes.** Three filters with counts: **All**, **Absent** (answered without you) and **Named**. Where Scout tagged the questions when it suggested them, more filters appear: the journey stage (Any stage, awareness, consideration, decision), the product the question is about, and the audience asking it. **A row, opened.** Collapsed, a question is one line with its status. Expanded, it shows the tone, **Named instead** (who came up in your place), **Built from** (the pages the assistant retrieved), **Cited** (the domains), **Sample** (how many runs it has been through) and the answer itself as plain prose. **Next steps.** In the Absent lane, a bar above the list reads: _"Each of these is a question a buyer asked and an assistant answered without you. Scout can research n as topics and carry the ones you approve to a draft."_ **Research these n** sends up to 25 at once. Each row also carries its own next step, one of three: the research action; **Next step: Sent to research** with the date, offered only once because a second send is a second bill; or a link to the opportunity the research produced. **Before there is a panel.** With no panel confirmed, the tab says **No prompt panel yet.** and explains what a panel is, with a **Build your panel** button. With a panel confirmed but never run, it says Scout is tracking your questions and asks them every morning, that the first answers land tomorrow, and offers **Run the panel now** and **Edit the panel**. ## Sources **What the answers were built from.** Who the assistants named instead of you, and the pages they trusted to answer. ![The Sources tab: who else was named and what the answers were built on](/docs/screens/ai-visibility-sources.png) _Being named is worth less than being the page the answer came from._ **Who else the assistants named.** The other companies that appeared in answers to your questions, counted only across your own panel so a brand outside your category cannot crowd the denominator. **What the answers were built on.** When the assistants searched the web while answering, this lists the pages they actually retrieved. Scout reads each page once a week to see whether it lists you, and each row says so: **Yours**, **Yes**, **No**, or **Not read yet**. A page that does not list you is one to earn a place on; one that does is a placement to improve. Both are moves you will see on the Overview. When the assistants did not search, the heading changes to **What the assistants recalled** and the caveat changes with it: these are domains named from training data, recollections rather than sources. They can be out of date, and publishing will not move them. Before a run has recorded pages, the tab says **Nothing to read yet.** ## Market **The wider market, estimated.** A modelled view across a third-party corpus. Broad, cheap, and not about your own questions. ![The Market tab with share of voice and the citation gap](/docs/screens/ai-visibility-market.png) _One large brand in your confirmed competitor set can dominate this share, and the page says so._ Three numbers: **Your share** with your rank among the confirmed competitors, **Mentions**, and **AI search volume (estimate)**. Under them, a share-of-voice leaderboard and **The citation gap**, which sources the engines cite for your topics and whether you are among them. **Measure it now** is an explicit button because measuring the market costs money with a third party, and a page must never do that on its own. It runs in the background and the page says so, filling in when the measurement lands. The caveats on this tab are the point of it. AI search volume is modelled, never a count of real questions. Coverage for one assistant is US and English only. The corpus is not your panel. If your workspace sits outside the covered market, a locale notice says so. If one confirmed competitor holds most of the denominator, a **crowded out by** warning names it and offers a one-click way to remove it from the set. And with no competitors confirmed, the tab says plainly: _"Share of voice is a share of something."_ Confirm the [competitors](/settings/competitors) you actually compete with and Scout will measure your place among them. --- # Costs and cadence URL: https://docs.audience.so/ai-visibility/costs-and-cadence > When the AI visibility panel runs on its own, which actions spend from the workspace's monthly budget, and what happens at the ceiling. AI visibility is the one part of Audience you spend against on a schedule, so this page is deliberately plain about what runs on its own, what only runs when you press a button, and what happens when the budget runs out. ## Every morning Once a panel is confirmed, Scout asks every question on it each morning, of every assistant it can reach. The Questions tab says as much before the first run: Scout is tracking your questions and asks them every morning, and the first answers land tomorrow. If you would rather not wait, **Run the panel now** asks them today. Daily rather than weekly is a deliberate choice. A single reading carries real variance, because answer engines do not repeat themselves exactly. Four readings a month cannot separate a genuine change in your visibility from the assistant phrasing itself differently; thirty can. This is why the Overview draws a weekly trend once two weeks of readings exist, and why every single result carries the "one sample" caveat. An assistant that cannot be reached on a given morning is skipped with a note rather than failing the run, and its tile on the Overview keeps the last good reading. ## The budget Every workspace has a monthly AI budget with a ceiling. The prompt panel page shows it as a line above the questions: how much of the ceiling has been spent so far. Near the ceiling the line turns amber and adds a sentence you should read as a promise: the panel stops asking once the ceiling is reached, and reports the shorter sample it took rather than counting the unasked questions as misses. That last part matters. A budget that quietly turned the unasked half of your panel into "Absent" would make your visibility look worse the less you spent. Audience never does that. When the ceiling stops a run, the Runs figure and each question's sample count reflect what was actually asked, and the rest wait for next month. The size of a panel is the main lever on cost: fifty questions a day costs more than twelve. The panel cap of 50 questions, and the fact that only confirmed questions are asked, are there to keep the spend proportional to what you chose to measure. [Budgets and limits](/automation/budgets-and-limits) describes the wider budget the panel draws from. ## What only runs when you ask Nothing on the AI visibility section spends money on arrival. Each of these is a button: - **Suggest questions**, which drafts a panel from your company profile. - **Run now** on the panel, and **Run the panel now** on the Questions tab. - Asking one assistant again from its tile on the Overview. - **Measure it now** on the Market tab, which pays a third party for the wider-market estimate. - **Research this** and its variants, which start research on a question you lost. Reading the report is free. Opening any tab, expanding an answer, filtering questions, and reading the sources cost nothing. ## Who can change it Configuring and running the panel, and measuring the market, is for owners and admins. Editors and viewers can read every screen and, if they are editors, send a lost question to research. See [Who does what](/start-here/who-does-what). --- # Your company profile URL: https://docs.audience.so/playbook/your-company-profile > What Scout believes about your company, where that belief comes from, and how to correct it or accept what a re-read of your website found. Every topic Scout proposes is judged against one document: its profile of your company. The Playbook's **Your company** tab shows that profile in full, under the heading **What Scout knows about you.** Nothing here is hidden, and everything is yours to change. ![The company profile in the Playbook](/docs/screens/playbook-company.png) _Every topic is judged against this. The version and its author are always shown._ ## What the profile holds | Field | What it is used for | | -------------------------- | ------------------------------------------------------------------------------------------------------ | | **Summary** | What the company does, in a paragraph. | | **Positioning** | How it wants to be seen against alternatives. | | **Primary market** | Who it sells into. | | **Brand voice** | How the writing should sound. Drafts are checked against it. | | **Products** | Each product with a sentence on what it does. Fit asks whether an article could lead to one of these. | | **Audiences** | Who the buyers are and what they are trying to get done. Fit asks whether the searcher is one of them. | | **Topics Scout looks for** | The core subjects. These seed what Scout searches around. | | **Industries** | The sectors the company serves. | ## Versions and who wrote them The line under the heading says which version this is and where it came from: "Version 1 · built by Scout from your website, 15 September" or "edited by Dana". Every change makes a new version, so the queue can always say which version a topic was judged against. Before any profile exists, the tab says: "Scout has not built your company profile yet. It is written from your website and documents when your workspace is set up. Until then, topics cannot be judged against your business." ## Direction The panel at the top of the profile is the one part a person states rather than Scout infers. It is covered on its own page: [Direction](/playbook/direction). ## Editing **Edit** opens the profile as a form, with the direction fields first, then Summary, Positioning, Primary market, Brand voice, and repeatable rows for Products (name and "What it does, in a sentence") and Audiences (who they are and what they are trying to get done). Core topics, industries and value propositions are one per line. Saving writes a new version attributed to you, and Scout re-judges every open topic against it. See [Direction and re-judging](/opportunities/direction-and-re-judging) for what that moves. Owners and admins can edit. Everyone can read. ## Re-read my site **Re-read my site** asks Scout to crawl your website again and compare what it finds with the current profile. The button's note says it plainly: "Crawl your website again and show what changed before anything moves." Nothing changes on its own. When the re-read finishes, one of two things appears. If something changed, a card headed **Your website says something new**: "Scout re-read it on 20 September and would change 3 things. Nothing moves until you say so." Each field is shown before and after, with additions and removals marked. **Keep mine** dismisses the proposal. **Accept as version N** makes it the live profile and re-judges the queue. If nothing changed: "Scout re-read your website on 20 September and found nothing that changes this profile." The page updates itself while a re-read is running, so you can leave it and come back. ## Related - [Direction](/playbook/direction) - [Add your knowledge](/setup/add-your-knowledge), the documents that feed the profile and the rules. --- # Direction URL: https://docs.audience.so/playbook/direction > The two sentences that outrank your website when Scout judges a topic, and what happens when you set them. Your website says what you sold last year. Direction is where you say what the business is doing now. It is the one part of the company profile that a person states rather than Scout infers, and Scout weighs a topic against it before anything it read on your site. ## The two fields Direction is two fields at the top of [Your company profile](/playbook/your-company-profile): - **Focusing on**: the product, market or theme the business is moving towards. "Deepfake detection for trust and safety teams at media platforms." - **Moving away from**: what the business is leaving, even where the website still sells it. "Consumer voice-changer apps." Before you set them the panel explains itself: "Your website says what you sold last year. If the business is moving, a new product, a market you are leaving, say so here. Scout judges topics against your direction before anything it read on the site." The button is **Tell Scout where you're heading**. The same invitation appears on Opportunities, under the heading, when no direction is set: **Wrong kind of topics? Tell Scout where you are heading**. ## Why it outranks the website Scout reads your site to build its picture of you, and a site is always a little behind the business. A product you are retiring still has a page; a market you entered last month has none. Direction is how you tell Scout the difference. A topic that serves what you are moving away from reads as low fit even while the page is live, and a topic that serves what you are focusing on is judged as squarely yours even before the site catches up. ## What happens when you save it Saving direction writes a new version of the company profile, attributed to you, and starts a re-judge: every open topic goes back in front of the judge with the new reading, and the queue is re-scored. Approved and rejected topics stay as they were. The provenance line under the Opportunities heading advances to the new version and reads "and the direction you set". See [Direction and re-judging](/opportunities/direction-and-re-judging) for exactly what moves. Owners and admins can set direction. ## Related - [Your company profile](/playbook/your-company-profile) - [Reading the queue](/opportunities/reading-the-queue) --- # Rules URL: https://docs.audience.so/playbook/rules > The editorial rules every draft is graded against, where they come from, and how to add, reword, switch off or retire one. The Playbook's **Rules** tab is the standard Scout writes to. Every draft is graded against these rules, and the review score on each draft reports how many it met as "Your playbook 86%". The line under the heading counts them: "24 in force, 3 turned off. Every draft is graded against these." ![The Rules tab of the Playbook](/docs/screens/playbook-rules.png) _Rules are grouped by what they govern. Each carries its origin and its history._ ## Where rules come from Three places, and each rule says which. - **The standard editorial rubric** Scout starts every workspace with: accuracy, sourcing, structure, metadata. - **Your documents.** Rules read from the brand guidelines, style guides and SOPs you gave Scout during setup or later under [Add your knowledge](/setup/add-your-knowledge). These carry a link to the source and the excerpt they were read from. - **Your edits.** Patterns Scout noticed in how reviewers corrected drafts, proposed under [Learned rules](/playbook/learned-rules) and accepted by a person. ## The eight categories | Category | Covers | | -------------------- | ---------------------------------------------------- | | **Accuracy** | What has to be true, and what has to be sourced. | | **Voice** | How your writing sounds when it is working. | | **ICP and audience** | Who the writing is for, and what they already know. | | **SEO and search** | How a piece earns the search it is written for. | | **Structure** | Shape, order and how a reader moves through a piece. | | **Metadata** | Titles, slugs and descriptions. | | **Linking** | Where a piece points, and why. | | **Your own rules** | Rules specific to this workspace. | ## What a rule card shows - The rule itself, written as an instruction. - Its **category** and **severity**. Severity decides what a failure does in review: a blocking rule stops a draft from being approved until the flag is dealt with; others raise a flag the reviewer weighs. - **Provenance**: where it came from, with a link to the source document and the excerpt it was read from. - **Confidence**, for rules Scout read or learned rather than was told. - An **on/off toggle**. A rule that is off stays in the list but is not graded. - **Edit** and **delete**. - **"N versions"**, a disclosure that opens every earlier wording: "v2 · in force · 12 March 2026 · Dana". ## Adding and rewording **Add a rule** opens a dialog with the guidance "Write it the way you would say it to a new writer on their first day." Its fields: - **Name**. - **The rule**, with the note "Specific enough that two reviewers would agree on whether a draft breaks it." - **Category** and **Severity**. - **Evaluation mode**: whether the rule is checked mechanically, by Scout reading the draft, or both. Editing an existing rule opens the same dialog, headed **Edit this rule**, with the note "Rewording a rule keeps the old version. This becomes v3." Nothing is overwritten; the versions disclosure shows what changed and when. Owners and admins can add, edit, toggle and delete rules. Everyone can read them. ## How rules are used in review On every draft, Scout checks each rule in force and raises a flag for each one the draft breaks. The review sidebar shows the result as two numbers, the generic rubric and **Your playbook**, and lists the flags under **What Scout found**. A reviewer applies a fix, dismisses the flag with a reason, or sends it back with the draft. Dismissing with a reason is how Scout learns a rule is being read too literally. See [Flags, comments and sources](/content/flags-comments-and-sources). ## If the tab is empty "Your playbook is empty. Scout starts every workspace with a standard editorial rubric and adds the rules it reads from your own style guides. If neither has happened yet, finish onboarding or add the first rule yourself." ## Related - [Learned rules](/playbook/learned-rules) - [Draft review](/content/draft-review) --- # Learned rules URL: https://docs.audience.so/playbook/learned-rules > Rules Scout proposes from the corrections your reviewers keep making, and how nothing applies until a person accepts it. When the same correction happens to more than one draft, it is probably a rule nobody has written down. The Playbook's **Learned** tab is where Scout proposes those rules. Each one carries the badge **Learned from your edits**, and none of them apply to your writing until you accept them. ![The Learned tab of the Playbook](/docs/screens/playbook-learned.png) _Each proposal shows its confidence and the edits it was learned from._ ## Where proposals come from As reviewers edit drafts, apply flags, dismiss flags with a reason and send drafts back with notes, Scout keeps a record of what changed. Once a week it reads the last two months of that record and looks for the same correction happening more than once: a term always replaced, a structure always added, a claim always cut. When a pattern is clear enough, it appears here as a proposed rule, with the excerpts it was learned from. Proposals arrive weekly rather than after every edit, which is why the tab does not change from day to day. ## What a proposal shows - The rule, written as an instruction, as it would appear under [Rules](/playbook/rules). - A **confidence** percentage. - **Why Scout suggests it**, a disclosure quoting the edits the pattern was read from. - Three actions: **Accept**, **Edit first**, **Not a rule**. The heading counts them: "3 things Scout thinks it has learned. Patterns Scout noticed in your edits. None of them apply to your writing until you accept them." ## Deciding **Accept** moves the rule into your playbook, in force from the next review. **Edit first** opens the rule dialog with the proposed wording filled in, so you can tighten it before it applies. **Not a rule** dismisses it. Scout stops proposing that pattern. Owners and admins decide. Everyone can read the proposals. ## The Decided section Below the open proposals, a disclosure headed "Learned rules you have already decided on (N)" lists what was accepted and what was dismissed, so the history of what Scout suggested is not lost when the queue is cleared. ## If there is nothing here "Scout has not learned anything yet. As you edit drafts, Scout looks for the same correction happening more than once. When it finds a pattern, it proposes a rule here for you to accept or turn down." The fastest way to fill it is to review drafts and say why when you dismiss a flag or send one back. ## Related - [Rules](/playbook/rules) - [Flags, comments and sources](/content/flags-comments-and-sources) --- # Topics Scout won't propose URL: https://docs.audience.so/playbook/topics-scout-wont-propose > Standing rules that keep whole themes out of your queue before they are scored, how to write one, and how to tell when one is too broad. Saying **Not worth it** to a topic makes Scout weigh topics like it lower. Sometimes that is not enough: a product line you are retiring, a theme you want no part of. A rule under the Playbook's **Topics Scout won't propose** tab removes those topics before they are scored, and stops them being proposed again. ![The exclusions tab of the Playbook](/docs/screens/playbook-exclusions.png) _Each rule shows how many topics it has removed, which is how you tell a rule is too broad._ ## How a rule matches Every rule is a pattern and a way of matching it. | How | What it keeps out | The screen's hint | | ----------------------------- | ---------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- | | **exactly this keyword** | Only that phrase. | Use it for a single term you never want to see. | | **anything mentioning this** | Any topic whose keyword contains the phrase. | Broad by design; check the count after a rebuild. | | **anything about this theme** | Any topic that is really about this subject, whatever words the search uses. | "Scout reads each candidate and keeps out the ones really about this, whatever words they use." | The first two are decided by matching words. The third needs Scout to read each candidate, so it takes effect through a re-judge of the queue rather than instantly. ## Writing a rule Two ways in. From the Playbook: the **Keep out** field takes the pattern, **How** picks the match mode with a live description of what it will do, and **Add rule** saves it. Matching open topics leave the queue straight away. From a rejection: in the **Not worth it** dialog on any opportunity, tick **Keep topics like this out from now on** and pick the match mode. The rule is created with the topic's keyword as its pattern. See [Deciding](/opportunities/deciding). Editors, admins and owners can add and remove rules, because a rule is a decision about the queue. ## What a rule shows - The **pattern** and the **match mode**. - Its **origin**: "From rejecting 'voice changer app' by Dana, 4 March", or that it was written here. - **"removed 12"**, or "removed none yet": how many topics this rule has taken out of the queue. - A one-click remove. The line under the heading counts them: "3 rules keep topics out of your queue." ## Reading the count The removed count is the number to watch. A rule you wrote for one theme that has removed sixty topics is probably matching more than you meant, especially if its mode is "anything mentioning this". Opportunities shows the same total under its heading, "N topics kept out by your rules", with a link back here, so a too-broad rule is always one click from being found. ## What a rule never touches Rules act on open topics: new, flagged and shortlisted. A topic someone has approved, rejected or is working on is a more specific decision than the rule and is left alone. Removing a rule does not bring back the topics it removed; they stay under **Not worth it** with the rule named as the reason, and Scout may propose them again on its next scan. ## Related - [Deciding](/opportunities/deciding) - [Direction](/playbook/direction), for a broader statement of what you are moving away from. --- # Reports overview URL: https://docs.audience.so/reports/overview > How the Reports page answers whether Audience is improving growth, what it shows before anything Scout wrote has gone live, and how to read the insights it raises. Reports has three tabs, and each is phrased as the question it answers: **Is Audience improving growth?**, **Which pages worked?** and **Is the operation getting more efficient?** The tab and the time window both live in the address, so a report you are looking at is a link you can send to someone else. ![The Reports overview](/docs/screens/reports-overview.png) _Every number is measured against the period before it._ ## The window The switcher at the top sets how many days the report covers. Whatever you pick, the comparison is the same length of time immediately before it, and the page says so: _"measured against the {n} days before"_. A 30-day report compares against the 30 days before that. ## Before anything has gone live Until an article Scout produced has been published, the Overview tab is in baseline mode. The headline becomes **Where your search performance stands**, and the subhead explains that nothing Scout produced has gone live yet and that this is the baseline Scout measures itself against. Changes are shown but rendered neutral, neither green nor red, because they are not yet anything Scout did. A sentence reads the baseline for you, in one of two directions: - _"Organic clicks are down 12% on the previous 30 days, that slide is the problem Scout is set up to reverse."_ - _"Organic clicks are up 8% on the previous 30 days, Scout's job is to compound that."_ Then: _"Everything below is the before picture."_ ## The metric cards - **Organic clicks** and **Impressions** from Search Console, each with its change. - **Page-one keywords**, with the date it was measured, or _"Scout takes the first ranking snapshot during setup."_ before it has. - **Published**, the pieces that went live in the window. In baseline mode this card is **Waiting on you** instead, the outlines and drafts awaiting a decision, with the note that approving them is what starts moving the numbers. ## Outcomes Clicks are the means; outcomes are the end. With Google Analytics connected, this section shows **Sessions**, **Key events** (the conversions your Analytics property counts) and **Influenced revenue** from the pages Scout can attribute. Where Analytics records no revenue metric, the card says _"No revenue metric configured"_ rather than showing a zero. Without Analytics the section reads **Outcomes are not measured yet.** and explains what connecting it would add. See [connect Google Analytics](/setup/connect-google-analytics). ## What you decided A small tally: insights accepted, and insights awaiting a decision. ## What fills in from here In baseline mode, Outcomes and What you decided are replaced by three short definitions, Outcomes, What you decided and What Scout noticed, each naming the thing that turns it on. Read it as the map of what the page becomes. ## What Scout has noticed Insights are findings Scout drew from your own numbers. Each shows its type, the date, a confidence, a title and a description, and its evidence figures inline rather than behind a click. Under each is a **Do next** recommendation, the content it covers, and two buttons: **Accept** and **Dismiss**. Accepting an insight changes what Scout plans next; dismissing one tells it the pattern does not matter here. Before there is enough published work to compare, the list says so: _"No findings yet. Scout looks for patterns across published work once there is enough of it to compare."_ Owners and admins accept and dismiss insights. See [who does what](/start-here/who-does-what). ## The summary card When a narrative summary is available, a card at the top of the tab gives the headline, a short summary and a few sections written from the same numbers. It is regenerated on a regular cadence rather than on every visit. ## Nothing to report yet Reports read Search Console data. Without it, the page says **Nothing to report yet.** and explains that it fills in on its own once Search Console is connected and the first sync lands. See [connect Search Console](/setup/connect-search-console). ## The other tabs [Content performance](/reports/content-performance) answers which pages worked. [Operations](/reports/operations) answers whether the operation is getting more efficient. --- # Content performance URL: https://docs.audience.so/reports/content-performance > The report that says which pages worked, what Scout recommends doing about each one, and how the rest of your site compares with the pages Scout published. **Which pages worked?** is the question this tab answers. It has two tables: the pages Scout published, each with a verdict on what to do next, and the rest of your site, where the fastest wins often are. ![Content performance](/docs/screens/reports-content.png) _Each page Scout published ends in a verdict under "Scout says"._ ## Pages Scout published Every article that went live, measured two ways at once: against the period before the report's window, and against the page's own first month. The table's subhead says so. | Column | What it shows | | ----------- | ------------------------------------------------------------------------------------ | | Page | The title, linking to the single-page report | | Published | When it went live | | Clicks | Organic clicks in the window | | Change | Against the period before | | Impressions | How often it appeared in search | | CTR | Clicks divided by impressions | | Position | Average position in search results | | Key events | Conversions the page produced. Shown only when Google Analytics is connected | | AI mentions | How often assistants mentioned the page's topic. Shown only when AI visibility is on | | Scout says | The verdict | ### Scout says The last column is a badge with a next move. Hover it for the reasoning. - **Maintain.** The page earns its place as it is. _"Nothing needs doing."_ - **Refresh.** The page is losing ground or its subject has moved on. Scout's next scan weighs it for the queue; nothing is needed from you yet. - **Expand cluster.** The page is working and the questions around it are uncovered. A cluster of related pages would compound it. - **Investigate.** The numbers moved in a way search data alone cannot explain. _"Worth a human look at the page itself."_ When a verdict has already become an opportunity, the badge turns green and links to it: _"In your queue as '{topic}', click to open."_ ### Before anything is published The table says _"Nothing published yet."_ and sets the expectation: once the first piece goes live, its performance appears here about a week later, because search takes that long to notice a new page. Without Analytics, a line invites you to connect it _"to see what these pages converted, not only what they attracted."_ ## The rest of the site Pages Scout did not write, from the same Search Console data. The table's own subhead makes the point: _"The fastest wins are usually here."_ A page that already ranks on the second page, or earns impressions without clicks, is often a smaller job than a new article. Scout lists the pages that earn enough impressions to be worth a decision; when none do, it says _"No other pages to compare against."_ ## The single-page report Click a page title to open its own report. It shows three comparison windows side by side, each answering a different question: 1. **Is it moving now?** The current window against the one before. 2. **Did it ever land?** The page's first month against the period before publication. 3. **Did publishing change anything?** After against before. Under them, a clicks series over time and a table of the queries the page appears for. A page that was published before Search Console history reaches back has no pre-publication baseline, and the report says so rather than showing a zero. ## The one-week lag Search Console data arrives two to three days behind, and a new page takes about a week to start appearing in results. A page published yesterday will show nothing yet, and that is not a verdict on the page. ## Related [Reports overview](/reports/overview) for the numbers across the whole site. [Operations](/reports/operations) for how efficiently the pages were made. --- # Operations URL: https://docs.audience.so/reports/operations > The report on whether the operation is getting more efficient, covering how you decide, how fast work moves, how quality trends, and where the friction is. **Is the operation getting more efficient?** This tab measures the process rather than the pages: how often you approve what Scout proposes, how long each stage takes, how draft quality is trending, and where effort and failures pile up. Every figure is for the report's window and compared with the period before. ![Operations](/docs/screens/reports-operations.png) _Decisions, speed, quality and friction, each with its sample size._ ## Decisions Three rates, each with its definition beside it and the sample it was measured on, for example _"Across 14 drafts"_. - **Opportunity approval rate.** Of the topics Scout proposed, the share you moved forward. - **First-pass approval rate.** Of the drafts you reviewed, the share approved without being sent back. - **Revisions per piece.** How many rounds a draft went through before approval. Before anything has been decided, the section says _"Nothing decided yet."_ ## Speed How long each hand-off takes: - **Opportunity to approved outline.** From moving a topic forward to approving its outline. - **Approved outline to draft.** How long Scout takes to produce the first draft. - **Editor time (estimate).** How much of a person's time the review took. - **Review sessions.** How many separate sittings a review needed. Editor time carries a note you should take at face value: _"Editor time is an estimate, not a measurement."_ Scout infers it from the timestamps of the actions people took and the changes they made. It cannot see a tab left open, a reading pass with no edits, or a discussion elsewhere. _"Read it as a trend, never as an hours figure."_ ## Quality The average **Scout quality score** of drafts, per week. A rising line means drafts arrive closer to approval; a falling one is worth looking into on the [Playbook](/playbook/rules), since the score is measured against your rules. ## Where the friction is Two views of where the work goes: - **Most failed categories.** Which playbook categories drafts fail most often. A category that keeps failing is either a rule Scout has not learned yet or a rule that needs rewording. - **Effort by stage.** Actions taken and estimated minutes per stage, so you can see whether time goes into outlines, drafts or flags. The estimate uses the same method as editor time; the tooltip explains it. ## The empty state Until a piece has been through review, the tab says _"The operation has not started yet."_ and tells you what starts it: approve an opportunity and take a piece through review, and the page begins reporting how long each stage takes and where the effort goes. ## Related [Reports overview](/reports/overview) for growth; [Content performance](/reports/content-performance) for the pages themselves. The [Activity page](/automation/activity) shows the same work as it happens. --- # What runs when URL: https://docs.audience.so/automation/what-runs-when > Everything Scout does on its own and on what schedule, the one cadence you control, and the actions that never happen without a person pressing a button. Audience works while you are not looking at it. Most of what Scout does happens on a schedule: syncing your search data overnight, re-scoring the queue, asking the AI visibility panel each morning, proposing rules once a week. This page lists all of it, so nothing that appears on Home or in Activity is a surprise. ## On a schedule | When | What Scout does | Why it matters | | ------------------ | -------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Every night | Syncs the latest Search Console data, then Analytics if it is connected. | Search figures always run two to three days behind, whatever tool reads them. | | Every night | Re-scores every open opportunity against the fresh search data. | A flagged topic whose demand climbed above the minimum clears its guardrail and returns to the queue on its own. Scores on open topics refresh daily whatever your rebuild cadence. | | On your cadence | Rebuilds the opportunity queue: reads the market again, regroups keywords into articles, and judges each group against your company profile. | The expensive path, and the one that finds new topics. See below. | | Weekly | Refreshes keyword demand and live rankings for topics that are open or in progress. | Scoped to active work, so a workspace with nothing shortlisted costs nothing here. | | Daily | Takes a performance snapshot of every page Scout published. | Feeds Content performance and the single-page reports. | | Every morning | Asks the confirmed AI visibility panel, for workspaces that have one. | The recurring spend the panel screen warns about. See [Costs and cadence](/ai-visibility/costs-and-cadence). | | Weekly | Reads the pages assistants cited, to see whether they list you. | Turns "cited a roundup" into "a roundup to earn a place on" or "a placement to improve". | | Weekly | Proposes learned playbook rules from the last two months of your edits. | Why the Learned tab fills weekly, not daily. See [Learned rules](/playbook/learned-rules). | | Weekly | Looks for patterns across your published work and writes them up as insights. | "What Scout has noticed" on Reports. | | Weekly and monthly | Writes the report narrative for the week and the month just finished. | The summary card at the top of the Reports overview. | | Every few minutes | Checks for runs that stopped without finishing and closes them. | Why a failed run never sits as "running" forever. See [Activity](/automation/activity). | ## The one cadence you control The rebuild is the only scheduled job with a setting, because it is the one that costs. In Workspace settings, **Rebuild opportunities** offers: **Every day**, **Every three days**, **Every week**, **Every two weeks**, or **Only when I ask**. The help text beside it is the whole trade-off: > A full rebuild reads the market again, so more often costs more. Scores on open opportunities refresh daily either way, and you can always research a topic on demand. Weekly is the default and suits most teams. Choose **Only when I ask** if you would rather trigger rebuilds yourself and rely on [Research a topic](/opportunities/research-a-topic) between them. See [Workspace and members](/settings/workspace-and-members) for where the setting lives. ## Nothing that spends money happens on arrival Opening a page never costs anything. Every action that spends is a button a person presses: - **Research a topic** and its variants on the AI visibility screens. - **Suggest questions**, **Run now** and **Measure it now** on AI visibility. - **Re-read my site** on the Playbook. - **Move forward** on an opportunity, which starts research and drafting. - **Approve outline & start drafting** and **Write it again** on the outline. - **Retry** on a failed run. And nothing happens optimistically. When you approve, reject, accept or dismiss, the screen waits for the server to confirm before it shows the change. ## While Scout works Screens that have something in flight poll themselves and stop when it settles: Home and Opportunities every few seconds while a scan runs, an article's record while Scout is drafting, the setup screen throughout the build. You can leave any of them. The sidebar indicator shows **Scout is working** wherever you are, and [Activity](/automation/activity) shows every run and every agent call as it happens. --- # Activity URL: https://docs.audience.so/automation/activity > How to see what Scout is doing right now and what it did this week, from the sidebar indicator to the ledger of every agent call. Scout does its work in the open. Every run it starts and every model call it makes is visible while it happens and afterwards, with what each one read, how long it took and what it produced. Activity is where that lives. ## The sidebar indicator At the foot of the sidebar on every page sits the activity indicator. While nothing is running it is a quiet row that says **Activity**. The moment something starts, it becomes a live card: a pulsing dot, **Scout is working**, the name of the run, a progress bar, **n agents on it**, and what the current agent is doing and to what ("Judging fit · deepfake detection api"). Completion toasts appear here too, so you learn that a scan finished without leaving the page you were on. Clicking the indicator opens the Activity page. ## The Activity page ![The Activity page with a run in progress](/docs/screens/activity.png) _The headline is the state. Everything underneath explains it._ The heading tells you the state in one line, and the sentence under it says what that means: - **Scout is working — 3 agents on it right now.** Every agent below is reading your material and writing back into your workspace. This page updates as they go. - **Nothing running right now.** Scout picks work up on its own: when your profile changes, when a draft needs review, on its weekly cadence. Sometimes the sentence explains that Scout is fetching and measuring, pulling your search history and the numbers behind it, before any model work starts. - **Nothing is running — one run stopped early.** A run stopped before it finished and nothing is working on it any more. Scout closes runs like that on its own, so this clears itself. No action is needed from you. Under the heading, four figures: **Running now**, **Last 24 hours** (agent calls), **Time worked** (model time) and **Read and written** (tokens). If any calls did not finish, the figure says so. ## Three sections **Running now.** A card per run: which agents are on it, the current step, how long it has been going. Empty, it says: _"All quiet. Nothing is queued or running. Accepting a company profile, asking for a draft or changing your rules will put agents to work here."_ **This week.** Every finished run from the past seven days with its outcome or its error, how many agent calls it took and how long. A failed run carries a **Retry** button. Retrying picks up exactly where the run stopped rather than starting over, so a scan that judged half its topics before failing only judges the other half. **Every agent call.** The newest sixty, whatever run they belonged to. Each one is a model reading your material and writing back. The columns are **When**, **Agent** (its role and the model), **Working on** (the activity and its subject), **Consulted** (what it read), **Took**, **Tokens** and **Status**. ## What to do with it Mostly nothing. Activity exists so you never have to wonder whether Scout is doing something, and so you can retry a run that failed without asking anyone. If a run has stopped early and the page says it will clear itself, it will. The same runs appear on Home under **Running now** and **What Scout did this week**, with the same Retry. For what Scout does on its own and when, see [What runs when](/automation/what-runs-when). For what it will not spend without asking, see [Budgets and limits](/automation/budgets-and-limits). --- # Budgets and limits URL: https://docs.audience.so/automation/budgets-and-limits > The monthly budgets, per-article limits and plan entitlements that keep Scout's spending inside what you agreed to, and what happens when one is reached. Scout spends money on your behalf: model calls to judge, research and write, and paid search data to measure. Every one of those spends is checked against a limit before it happens, not reported after. When a limit is reached, Scout stops and says so. It never spends past the line and it never hides the fact that it stopped. ## The monthly budgets Each workspace has two monthly ceilings: one for AI (the model work behind judging, research, drafting and review, and the AI visibility panel) and one for paid data (search demand, rankings, live search results). A warning appears when either reaches 80% of its ceiling. The prompt panel page shows the AI budget as a line above the questions, and turns it amber near the ceiling. At the ceiling, scheduled work that would spend stops until the next month begins. What has already been measured stays exactly as it was. The AI visibility panel is the clearest example: it stops asking once the ceiling is reached and reports the shorter sample it took, rather than counting the questions it never asked as misses. See [Costs and cadence](/ai-visibility/costs-and-cadence). ## Per-article limits A single article cannot run away with the budget. Each piece has a ceiling for everything Scout spends on it across research, outline, draft, review and revisions, and a cap on how many separate agent runs it may take. Each individual run has its own limits too: how many steps it may take, how many tools it may call, how long it may work. A run that reaches one of them stops and the article is marked as needing attention, with a **Try again** button on the Content page. ## Plan entitlements The plan sets what a workspace may hold. These are the values shown in the product today. | Entitlement | Limit | | ------------------------------------ | ----- | | Sites per workspace | 1 | | Pages crawled from your site | 300 | | Generated articles per month | 30 | | Live search result checks per month | 500 | | Members per workspace | 10 | | Documents in the playbook | 50 | | Size of one uploaded document | 50 MB | | Questions on the AI visibility panel | 50 | Where a limit is reached, the screen that would take you past it says so in place. The knowledge step of setup, for example, reads _"This workspace is holding its limit of 50 documents. Remove one to add another."_ ## Manual refreshes Some things can be asked for by hand and cost money each time, so they are rate-limited per day: | Action | Limit | | ------------------------------------------- | ------- | | Opportunity refresh (a rebuild you trigger) | 3 a day | | Re-read my site (a fresh crawl) | 1 a day | | Search data refresh | 2 a day | When you reach one, the button explains when it will be available again. Scheduled work is not affected by these limits. ## What "stopping" looks like Three rules hold everywhere a limit applies: 1. **Scout stops before spending, not after.** The check happens before the call. 2. **It reports the shorter sample.** A panel that asked 30 of 50 questions says 30. A scan that judged half its topics says which half. 3. **It never stops silently.** The run appears on [Activity](/automation/activity) and on Home with its reason, and anything that can be retried carries a **Retry** button. Owners can see all of this on the workspace, and it is owners and admins who decide the rebuild cadence in [Workspace and members](/settings/workspace-and-members), the setting with the biggest effect on monthly spend. --- # Workspace and members URL: https://docs.audience.so/settings/workspace-and-members > Managing your workspaces, their names and rebuild cadence, who belongs to them and in what role, and how to invite, remove and switch. A workspace is one company, one site and one team. Everything in Audience belongs to a workspace, and every address in the product starts with the workspace you are in. You can belong to several, and switch between them from the sidebar. ## Your workspaces Open the account menu and choose **Workspaces**, or go to Settings → Workspaces. The list shows every workspace you belong to: its name, a **Personal** badge where it is yours alone, and your role in it. Each row offers what your role allows: **Edit workspace** for owners and admins, **View workspace** for editors and viewers, and **Leave workspace** for anyone who is not the owner of a shared workspace. ![The workspaces list](/docs/screens/workspaces.png) _Your role in each workspace decides which buttons appear._ **New workspace** starts a fresh one and takes you into [setup](/setup/set-up-your-workspace). ## Workspace settings Open a workspace's settings from the list, or from the workspace menu in the sidebar (**Workspace**). Owners see the settings block at the top; everyone else sees the workspace name as a heading. ![Workspace settings with members and pending invitations](/docs/screens/settings-workspace.png) _Name, rebuild cadence, members, invitations, and, for owners, deletion._ **Workspace name.** What appears in the switcher and at the top of every invitation. **Rebuild opportunities.** How often Scout reads the market again to find new topics: **Every day**, **Every three days**, **Every week**, **Every two weeks** or **Only when I ask**. The help text explains the trade-off: a full rebuild reads the market again, search results, clustering and one judgement per topic group, so more often costs more; scores on open opportunities refresh daily either way, and you can always research a topic on demand. See [What runs when](/automation/what-runs-when). ## Members **Workspace members** lists everyone in the workspace with their avatar, name, email and a role dropdown. Change a role by picking a new one; the change saves at once. Remove a member with the button at the end of their row. The owner's row is fixed: an owner's role cannot be changed here and an owner cannot be removed or leave. The roles are **Admin**, **Editor** and **Viewer** when you invite or reassign; the owner is whoever created the workspace. [Who does what](/start-here/who-does-what) lists what each role can do. ## Inviting people **Invite member** asks for an email address and a role. The person receives an email with a link. Someone who already has an Audience account accepts from Home, where a **Pending workspace invitations** dialog lists every invitation waiting for them with **Accept** and **Decline**. Someone new creates their account from the link, because accounts in Audience are created by invitation only. Until it is accepted, an invitation sits under **Pending invitations** with its email and role and a cancel button. Invitations that are never accepted expire on their own. [Invite your team](/setup/invite-your-team) walks through it. ## Deleting a workspace At the bottom of the settings, owners of a shared workspace see **Delete workspace**. The warning is plain: this cannot be undone. Everything the workspace holds, including its opportunities, articles, playbook and reports, goes with it. A personal workspace cannot be deleted from here. ## Switching workspaces The switcher at the top of the sidebar shows the current workspace's name with its site domain underneath. Choose another workspace and you land on the same kind of page you were on: from Opportunities to Opportunities, from Reports to Reports. Anything specific to the workspace you left, such as an open article, is dropped, since it would not exist in the next one. The switcher also holds the shortcuts to **Workspace**, **Integrations** and **Competitors** settings. --- # Integrations URL: https://docs.audience.so/settings/integrations > The Integrations settings page, where you connect, re-check or disconnect Google Search Console and Google Analytics after setup. **Integrations** is what Scout reads to find the work worth doing. During setup you connect them on step 2; afterwards this page, reached from the workspace menu in the sidebar, is the only way back. Only owners and admins can change a connection. ![The Integrations settings page](/docs/screens/settings-integrations.png) _The same two cards as setup, with the controls you need once a connection exists._ ## The two connections **Google Search Console**, marked **Strongly recommended**. What people already search to find you, and which of your pages are slipping. This is the single strongest signal Scout has, and it costs nothing to connect. [Connect Search Console](/setup/connect-search-console) explains what it unlocks and what Scout does without it. **Google Analytics**, marked **Optional**. What happens after the click, so Scout can report signups and revenue instead of traffic. [Connect Google Analytics](/setup/connect-google-analytics) covers the detail. ## What a card can say Each card shows the state of its connection and the one action that fits it. | State | What you see | | ----------------------------- | ----------------------------------------------------------------------------------------------------------------------- | | Not connected | A connect button. Pressing it opens Google's consent screen in the same tab and brings you back here when you are done. | | Started but not finished | **Continue connecting**. You left Google's screen before granting access. | | Connected | The property Scout is reading, as **Reading {property}.**, with **Sync now** and **Disconnect**. | | Expired, failed or revoked | **Reconnect**. The grant stopped working, usually because access was withdrawn in Google. | | Connected, no property chosen | **Choose a property**. Google returned more than one and the match was not obvious. Nothing is read until you pick. | **Sync now** re-checks the grant and pulls the latest data without waiting for tonight's sync. It is not a full re-import. ## Choosing a property When your Google account has exactly one property that matches your site, Scout picks it. When it does not, a picker lists what Google returned so you choose. Until you do, Home shows the connection as waiting on you, and setup's build screen marks the task **Waiting on you** rather than failed. ## Disconnecting Disconnecting stops Scout reading new data. The history it has already pulled stays where it is, so reconnecting the same property picks up where it left off. Nothing in your reports or opportunity scores is deleted; the parts that need fresh data simply stop updating and say so. ## Related - [Connect Search Console](/setup/connect-search-console) - [Connect Google Analytics](/setup/connect-google-analytics) - [What runs when](/automation/what-runs-when) for the nightly syncs. --- # Competitors URL: https://docs.audience.so/settings/competitors > The Competitors settings page, where you keep the set Scout measures you against accurate, because a wrong name quietly distorts three things at once. **Competitors** is who Scout measures you against. This set decides your share of voice, which gaps Scout proposes, and part of every opportunity score, so a name that does not belong here quietly distorts all three. The page is reached from the workspace menu in the sidebar. Only owners and admins can change it. ![The Competitors settings page](/docs/screens/settings-competitors.png) _The whole list, not only the strongest matches, because someone who came here is usually looking for one wrong name._ ## Where the list comes from Scout proposes competitors during [setup](/setup/confirm-competitors) by looking at who else ranks for the queries you already rank for, and it keeps proposing new ones as your search market shifts. Anyone you confirm or add stays. Anyone you remove is gone until you add them back. ## Classification Each competitor has a type, chosen from the dropdown on its row. The three are defined inside the dropdown itself: | Classification | Meaning | | ----------------------- | ---------------------------------------------------------------------------------------------------- | | **Direct competitor** | Sells something a buyer would weigh against you. | | **Search competitor** | Ranks for your topics without selling what you sell. | | **Reference publisher** | Scout cites these rather than trying to out-rank them: a publication, a directory, a standards body. | Edits save themselves. There is no Save button on this page. ## Why a wrong name matters The set feeds three things: - **Share of voice** on the AI visibility Market tab is a share of the mentions among this set. One large brand that does not belong turns everyone else's share into a rounding error, and the tab will warn you that you are being crowded out by it. - **Competitor gaps** in the opportunity queue are the searches a confirmed competitor is strong for and you are not. A reference site listed as a competitor fills the queue with gaps you could never close. - **Competitive position** is one of the eight components of every opportunity's [score](/opportunities/the-score). With no confirmed competitors it scores nothing, and the product says it does not know who you compete with yet. If a competitor is a common word as well as a company, Scout takes care not to treat that word as a brand when it filters your keywords; you do not have to do anything for that. ## Adding one you know Type a domain into the add field and choose its type. Scout starts measuring against it on the next rebuild and the next AI visibility market measurement. Removing a competitor is the button at the end of its row. The empty state says it plainly: _"No competitors yet. Scout proposes these from who else ranks for your topics. Add anyone you know you are compared against in the meantime."_ --- # Publishing destinations URL: https://docs.audience.so/settings/publishing-destinations > Where finished articles go, why publishing to a CMS is off in this release, and how an approved article leaves Audience as a Markdown file. **Publishing** is where finished articles go. Scout always writes a draft first; nothing is published until you say so. In this release, publishing straight to a CMS is switched off, and the page says so. Approved articles leave Audience as a Markdown download, and putting them live is a step you take in your own CMS. ![The Publishing settings page](/docs/screens/settings-publishing.png) _The page exists so that the decision is visible, not hidden. It offers no destination to add while CMS publishing is off._ ## What the page shows Open it from the workspace menu in the sidebar. With CMS publishing off, the page shows no destinations and explains that publishing straight to your CMS is off in this release. There is nothing to configure. If a destination had been set up in an earlier release, it would still be listed here with its name and site, but it would not be used. Only owners and admins can change where content is published. Everyone else sees the page and a line saying so. ## How an article leaves Audience When a draft is approved, the draft review screen offers **Download Markdown**, with the line: _"Approved and ready. Publishing is manual — put it live yourself."_ The file holds the approved version of the article with its title, meta description and slug, ready to paste into your CMS. [Publishing and Markdown](/content/publishing-and-markdown) walks through the download and a short checklist for putting it live. Once the page is live on your site, Scout finds it in your search data and starts reporting on it under [Content performance](/reports/content-performance), usually about a week after publication. ## Why this is a decision, not a gap Auto-publishing is off in this release on purpose. A draft that can go live without a person confirming it in the CMS where it will appear is a draft that can go live wrong, and this product's whole shape is a human checkpoint at every decision that matters. The Publishing page is here so that the choice is stated in the product rather than left for you to discover. --- # Account and security URL: https://docs.audience.so/settings/account-and-security > Your profile, password, two-factor authentication, passkeys and appearance, and the dialog that shows workspace invitations waiting for you. Your account is yours across every workspace you belong to. These settings are reached from the account menu at the foot of the sidebar and are not tied to any one workspace. ![Profile settings](/docs/screens/settings-profile.png) _Name and email, and the notice you see until an address is verified._ ## Profile **Profile** holds your name and email address. Changing your email sends a verification link to the new address, and until you open it the page shows an unverified-email notice with a link to send it again. **Save** applies the change. At the bottom of the page is **Delete account**. It removes your account and your personal workspace. It does not delete shared workspaces you belong to; if you own one, transfer it or delete it first from [Workspace and members](/settings/workspace-and-members). ## Security **Security** asks you to confirm your password before it opens, then offers three things. - **Update password.** Your current password, then the new one twice. - **Two-factor authentication.** Turn it on to require a code from an authenticator app at login, and save the recovery codes it shows you somewhere safe. When you sign in with two-factor on, the login screen asks for the code, or a recovery code if you have lost the app. - **Passkeys.** Register a passkey to sign in with your device's fingerprint, face or PIN instead of a password. You can register more than one and remove any of them. ## Appearance **Appearance** sets the theme: **Light**, **Dark** or **System**, which follows your operating system. The choice is remembered on this device and also applies to the documentation you are reading now. ## Invitations waiting for you When a workspace owner or admin invites you and you already have an account, the invitation appears as a **Pending workspace invitations** dialog the next time you open Home. Each one shows the workspace name and who invited you, with **Accept** and **Decline**. Accepting adds the workspace to your switcher and takes you into it. Declining removes the invitation; the owner can send another. [Invite your team](/setup/invite-your-team) describes the other side of this. ## Signing in The front door is **Log in**. Accounts are created by invitation only, so there is no sign-up form unless you arrived from an invitation link. **Forgot your password?** sends a reset link to the address you signed up with. If your email is unverified, you are asked to check your inbox before you can go further. --- # Glossary URL: https://docs.audience.so/reference/glossary > Every term the Audience interface uses, in alphabetical order, with a sentence or two on what it means and where it appears. The words on Audience's screens, alphabetically. Where a term has its own page, the entry links to it. **Absent.** A question on the prompt panel that an assistant answered without naming you. The opposite of Named. Absent questions are the ones that carry a next step. See [Questions, sources and market](/ai-visibility/questions-sources-and-market). **Activity.** The record of everything Scout is doing and has done: runs, agent calls, how long each took, what it consulted. Reached from the sidebar indicator, which reads "Scout is working" while anything runs. See [Activity](/automation/activity). **Admin.** A role that can do everything an owner can except delete the workspace. See [Who does what](/start-here/who-does-what). **All signals.** The unfiltered state of the type filter on the opportunity queue. **Angle.** The specific way into a topic Scout recommends: not "write about X" but why this article, for this reader, from this direction. Shown in serif on each opportunity. **Approve & move forward.** The button on draft review that marks the current version approved. Nothing is published by it. **Approve outline & start drafting.** The button on outline review. What you have saved becomes the version Scout drafts from. **Approved.** An opportunity you moved forward, or a content version you accepted. **Blocked.** A review routing: a finding is stopping approval and has opened itself. The routing line names the actual blocking flag. **Build.** The final setup step, where Scout assembles the workspace: profile, playbook, keywords, market snapshot, opportunities. See [The build](/setup/the-build). **Company profile.** What Scout believes about your business, written from your website and editable by you. Every topic is judged against it. See [Your company profile](/playbook/your-company-profile). **Content item.** An approved opportunity moving through research, outline, draft and review. See [The pipeline](/content/the-pipeline). **Direction.** The part of the company profile a person states: **Focusing on** and **Moving away from**. It outranks what Scout read on the website. See [Direction](/playbook/direction). **Do it anyway.** The override on a flagged topic. Puts it in the queue with your reason recorded. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). **Download Markdown.** How an approved article leaves Audience in this release. See [Publishing and Markdown](/content/publishing-and-markdown). **Dropped.** A prompt-panel lane: questions taken off the panel. Not asked, not re-proposed, restorable. **Editor.** A role that decides topics, starts research and approves drafts, but does not manage settings or publish. See [Who does what](/start-here/who-does-what). **Find another angle.** Asks Scout to propose a different angle for an open opportunity. **Flagged.** An opportunity that failed a guardrail. Kept out of the queue and out of the bin, with the reason attached. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). **Guardrail.** One of four thresholds a topic must clear: minimum search volume, maximum keyword difficulty, minimum score, minimum fit with the business. Each is reported separately. **Home.** The weekly briefing: what to move forward, what is waiting on you, whether it is working, what Scout learned. See [Your weekly briefing](/home/your-weekly-briefing). **In progress.** An opportunity that has become a content item and is being worked on. **Insight.** A pattern Scout found in your published work's numbers, with its evidence and a **Do next**. Accepted or dismissed by a person. See [Reports overview](/reports/overview). **Keep topics like this out from now on.** The option in the rejection dialog that creates a topic exclusion. **Lane.** One of the signals Scout finds topics from: what you nearly rank for, pages slipping, competitor gaps, expansions around what you sell, and pages of yours that already work. The queue records which lane found each topic. **Learned rule.** A rule Scout proposes after noticing the same correction in your edits more than once. Nothing applies until you choose **Accept**, **Edit first** or **Not a rule**. See [Learned rules](/playbook/learned-rules). **Market.** The AI visibility tab that estimates share of voice across a third-party corpus rather than your own questions. Measured on request because it spends with an outside provider. **Move forward.** Approves an opportunity and starts research. See [Deciding](/opportunities/deciding). **Named.** A question on the prompt panel that an assistant answered by naming you. Also **Named instead**: who came up when you did not. **Needs revision.** A review routing: worth another pass before this goes out. **New page, Refresh, Slipping, Nearly there, Gap, Cluster, AI visibility.** The type badges on the queue and on Home, saying what kind of work a topic is. **Not measured.** A score component Scout had no data for. Its weight moves to what could be measured; it is never counted as zero. **Not worth it.** Rejects an opportunity, with an optional reason Scout reads. Also the filter label for rejected topics. See [Deciding](/opportunities/deciding). **Open.** The default queue filter: New and Shortlisted topics. **Opportunity.** A topic Scout thinks is worth writing about, with its score, title, angle and reasoning. See [Reading the queue](/opportunities/reading-the-queue). **Outline review.** The stage where you approve the thinking before the writing. See [Topic and outline review](/content/topic-and-outline-review). **Owner.** The role that created the workspace. Can do everything, including delete it. **Playbook.** The rules Scout writes to, your company profile, what it has learned, and what it will never propose. See [Rules](/playbook/rules). **Prompt panel.** The buyer questions Scout asks AI assistants every day. You confirm it; nothing is asked until you do. See [Build your prompt panel](/ai-visibility/build-your-prompt-panel). **Quality score.** Scout's reading of a draft, out of 100, from a generic rubric and your playbook. Shown as **Scout quality score** on draft review. **Ready for review.** A review routing: nothing is blocking this. Read it, then move it forward. **Rebuild opportunities.** The workspace setting for how often Scout reads the market again: every day, every three days, every week, every two weeks, or only when you ask. See [What runs when](/automation/what-runs-when). **Re-read my site.** Asks Scout to crawl your website again and propose profile changes. Nothing changes until you accept. See [Your company profile](/playbook/your-company-profile). **Research a topic.** Give Scout a topic, a search term, a keyword list or a competitor, and it finds what people search around it. See [Research a topic](/opportunities/research-a-topic). **Runs.** On the AI visibility scorecard, how many times the panel has been asked. Each run is one sample, not a census. **Scout.** The agent inside Audience. Scout reads, finds, judges, researches, drafts and checks. You decide. **Send back.** Returns a draft to Scout with your notes and any findings you ticked. **Set aside.** Candidates Scout removed before judging because they were plainly not about your business. The queue shows the count and the most searched examples. See [Topics Scout set aside](/opportunities/topics-scout-set-aside). **Share of voice.** On the prompt panel, the share of your questions an assistant answered by naming you. On Market, an estimate across a wider corpus. **Shortlisted.** An open opportunity a person kept aside. Its score still refreshes, its status does not change on its own. **Site.** The website a workspace is about. See [Your website](/setup/your-website). **Suggested.** A prompt-panel lane: questions Scout proposed that you have not kept yet. Nothing there is asked or paid for. **This week's moves.** The AI visibility overview's list of the losses with the clearest fix, grouped by kind. See [This week's moves](/ai-visibility/this-weeks-moves). **Tone.** How an assistant described you, from -100 to 100. Blank when nobody assessed it; blank is not neutral. **Topics Scout won't propose.** The playbook tab for topic exclusions: rules that remove topics before they are scored. See [Topics Scout won't propose](/playbook/topics-scout-wont-propose). **Version.** One outline or one draft, marked as written by Scout or by a person. Nothing is overwritten. See [Versions and approval](/content/versions-and-approval). **Viewer.** A role that reads everything and can comment on drafts. **Waiting on you.** A content stage that needs a person: topic review, outline review or draft review. The pipeline shows how long each has waited. **Where you stand.** The AI visibility scorecard: questions you appear in, share of voice, tone, position and runs. See [Where you stand](/ai-visibility/where-you-stand). **Workspace.** One company's space in Audience: site, profile, playbook, queue and people. See [Workspace and members](/settings/workspace-and-members). **Write it again.** Asks Scout for a different outline. The current one is kept. **Your panel.** The prompt-panel lane of questions asked on every run. --- # Frequently asked questions URL: https://docs.audience.so/reference/faq > The questions marketing leads ask most about Audience, answered plainly, with links to the pages that go deeper. Short answers to the questions people ask in their first weeks with Audience. Each links to the page that covers the subject properly. ## Why is my queue empty? Three common reasons. Setup has not finished: Home shows a build-in-progress panel until it has, and the first scan runs once your site is crawled. Everything Scout found is flagged: the queue then says so and offers a button to the flagged pile. Or Search Console is not connected and your competitors are not confirmed, in which case Scout has little to work from. The empty state on the queue tells you which. See [Reading the queue](/opportunities/reading-the-queue). ## Why is everything flagged? Because every topic Scout found failed a guardrail, most often fit with what you sell. That usually means the topics your market's search data produced are not about your business, and Scout is being honest about it. Open the **Flagged** filter to read each reason, use **Do it anyway** on any you want, and use **Research a topic** to hand Scout topics you know matter. If the profile Scout built is wrong, fix it and every topic is judged again. See [Guardrails and flagged topics](/opportunities/guardrails-and-flagged-topics). ## Why does a score component say "not measured"? Scout had no data for it: Search Console is not connected, no competitors are confirmed, or a provider did not return a figure. Rather than counting it as zero, Scout moves that component's weight to the ones it could measure and says so. The score is still out of 100; it rests on less. See [The score](/opportunities/the-score). ## Can Scout publish for me? Not in this release. An approved article leaves Audience as a Markdown download and you put it live in your own CMS. The Publishing settings page says publishing straight to a CMS is off. Nothing Scout writes becomes public without a person doing it. See [Publishing and Markdown](/content/publishing-and-markdown). ## Why did my opportunities change after I edited the company profile? Every open topic is judged against the profile. When the profile or the direction changes, Scout re-judges every open topic against the new version, so scores move and some topics change status. Decided topics are left alone: anything approved, rejected or in progress keeps the score it was judged on. See [Direction and re-judging](/opportunities/direction-and-re-judging). ## How do I stop Scout proposing a theme? Reject one topic of that kind with **Not worth it**, tick **Keep topics like this out from now on**, and choose how to match: exactly this keyword, anything mentioning it, or anything about this theme. Matching open topics leave the queue at once and the rule appears in the playbook under **Topics Scout won't propose**, with a count of what it removed. See [Topics Scout won't propose](/playbook/topics-scout-wont-propose). ## Why is Search Console "strongly recommended"? It is the single strongest signal Scout has and it costs nothing. It tells Scout which queries already bring people to you, which pages are slipping and which are nearly on page one. Without it, Scout reasons from market data alone, part of every score is marked not measured, and the Reports overview has nothing to report. See [Connect Search Console](/setup/connect-search-console). ## What does "one sample" mean on AI visibility? Each time the prompt panel runs, every question is asked of each assistant once. Answer engines do not repeat themselves exactly, so a single answer is one reading, not the truth. Read the pattern across runs rather than any one line; the trend appears once there are two weeks of readings. See [Where you stand](/ai-visibility/where-you-stand). ## How much does the AI visibility panel cost? The panel screen shows what it costs in the only terms that matter: how much of this workspace's monthly AI budget is spent, against the ceiling. Near the ceiling the line turns amber and warns that the panel stops asking once it is reached, and reports the shorter sample it took rather than counting unasked questions as misses. Nothing is asked or paid for until you confirm the panel. See [Costs and cadence](/ai-visibility/costs-and-cadence). ## Can I undo a rejection? Not in this release. A rejected topic stays under the **Not worth it** filter with the reason you gave, and Scout never reverses a decision a person made. If you created an exclusion rule at the same time, removing the rule in the playbook stops it applying to new topics, but the ones it already removed stay where they are. See [Deciding](/opportunities/deciding). ## Why does a draft say "Blocked"? A finding with blocker severity is stopping approval, most often claims that could not be traced to a source Scout read. The routing line names the blocking flag and it opens itself in the findings rail. Apply the suggestion, dismiss it with a reason, or tick it and send the draft back. See [Flags, comments and sources](/content/flags-comments-and-sources). ## Does Scout change my draft when it finds a problem? No. Scout reports what it found and why, and nothing else happens until you decide. Applying a finding is your action; dismissing one with a reason teaches Scout to stop raising it. See [Draft review](/content/draft-review). ## What happens to my edits? They are saved as a new version, marked with your name, alongside Scout's. Nothing is overwritten, and the approved version is the one marked approved. See [Versions and approval](/content/versions-and-approval). ## Scout re-read my site. Did it change my profile? No. A re-read writes a proposal that shows every field it would change, before and after. Nothing moves until you press **Accept as version N**; **Keep mine** discards it. See [Your company profile](/playbook/your-company-profile). ## What does Scout do on its own? Scores refresh daily against new search data. The full rebuild runs on the cadence you set for the workspace. The prompt panel, once confirmed, runs every morning. Once a week Scout proposes rules it has learned from your edits. Everything else that spends against your budget is a button. See [What runs when](/automation/what-runs-when). ## Why can I approve a draft but not publish it? Because those are different decisions with different weight. Editors decide topics and drafts; putting something live belongs to owners and admins. In this release that act is the Markdown download. See [Who does what](/start-here/who-does-what). ## Why does the queue say Scout set aside some topics? Before judging, Scout removes candidates that are plainly not about your business, so search volume alone cannot fill the queue with topics that have nothing to do with you. The queue shows how many, and the most searched examples, so you can check. If one belongs to you, research it by name and Scout will judge it properly. See [Topics Scout set aside](/opportunities/topics-scout-set-aside). ## Can I add my own topics? Yes. **Research a topic** takes a topic, a search term, a pasted list of keywords, or a competitor's domain, and Scout finds what people search around it. The results are judged and scored like everything else. Editors and above can start research, because it spends against the workspace budget. See [Research a topic](/opportunities/research-a-topic). ## Why is Analytics optional if Search Console is not? Search Console is how Scout finds work. Analytics is how it reports what the work produced: sessions, key events and revenue instead of clicks. Without it, Reports say outcomes are not measured yet and Home invites you to connect it. It is worth connecting, but Scout can find topics without it. See [Connect Google Analytics](/setup/connect-google-analytics). ## Where do I see what Scout is doing right now? The sidebar indicator turns into a live card whenever anything runs, and the Activity page has every run and every agent call. A run that stopped early shows a **Retry** button. See [Activity](/automation/activity). --- # Reading these docs with a language model URL: https://docs.audience.so/reference/for-language-models > How to point an assistant at the Audience documentation: the llms.txt index, the single-file version, Markdown for every page, and stable section links. These docs are written to be read by people and by the assistants people ask. Every page exists as plain Markdown, the whole site is indexed in one file, and every heading has a stable address, so an assistant can find, quote and cite a section exactly. ## The index: /llms.txt [/llms.txt](/llms.txt) follows the llmstxt.org convention. It opens with a one-paragraph summary of what Audience is, then lists every page by section with its title, its Markdown address and its one-sentence description. Give an assistant this file first when you want it to decide which pages to read. ## Everything in one file: /llms-full.txt [/llms-full.txt](/llms-full.txt) is every page, in reading order, in one plain-text file. Each page begins with its title, its address and its description, so a quote can be traced back to the page it came from. Use it when you want an assistant to have all of the documentation in context at once. ## Any page as Markdown Add `.md` to any page's address and you get the source as written, with the title and description at the top. For example, the page at `/opportunities/the-score` is also at `/opportunities/the-score.md`. Every page links to its own Markdown twin at the bottom, and declares it in the page head as an alternate representation. ## Section links that do not move Every second- and third-level heading has an id made from its text, lowercased and hyphenated, and that id is the link target. A heading called "What stays with you" is reachable at `#what-stays-with-you` on its page. The ids survive edits to the surrounding prose, so a citation to a section keeps working. To cite a section, give its page address and the heading id together. ## What the docs cover The documentation describes the product as a person uses it: what each screen shows, what each button does, what happens next, and which decisions belong to which role. It covers setup, the opportunity queue and its score, the content pipeline from research to approval, AI visibility and its prompt panel, the playbook, reports, what runs automatically, and the settings. ## What the docs do not cover Audience has no public API in this release, so there is no API reference. The docs do not describe the data providers or models behind the product, do not give per-run prices, and do not describe how the product is built. Where a number on a screen is an estimate or a single sample, the docs say so, and an assistant reading them should carry that qualification into its answer. ## The words that matter The product is **Audience**. The agent inside it is **Scout**. Interface labels are quoted exactly as they appear, and the [glossary](/reference/glossary) defines every one of them. When an assistant answers a question about Audience, using those words as the interface does will make its answer easier for a person to act on.