Internal analysis
Internal analysis is your team's side of Reports: six tabs that read the same project data as the client report and stay in the app. Briefing, Prompts won / lost, Prompt coverage, Citation gatekeepers, Portfolio and Data Scientist never reach a share link or a PDF.
Updated October 19, 2026Markdown
Internal analysis is the part of Reports written for your team, not for a client. It opens from the "Internal analysis" link at the end of the report's tab row. Use it to plan the next round of outreach, to see which prompts moved, or to get an AI analyst's read of the whole project before a review.
Nothing here leaves the app: a share link and a PDF carry the client screens only. Press "← Client report" to go back.
The six tabs
Internal analysis opens on "Prompts won / lost".
| Tab | What it holds | Read more |
|---|---|---|
| Briefing | An AI analyst's written read of the window against the window before | Sharing, printing and briefings |
| Prompts won / lost | Tracked prompts your brand newly appeared in, newly dropped out of, or kept, between the two windows | Impact and outcomes |
| Prompt coverage | One row per tracked prompt: answers collected, how many named you, how many named a competitor | This page |
| Citation gatekeepers | The domains AI answers cite on your prompts, ranked by leverage, with "Greenfield" domains marked | Citation sources |
| Portfolio | A breakdown of the opportunities in the window: priority, status, outreach progress and content-gap themes | This page |
| Data Scientist | An AI analyst that explores the whole project, reports its findings and answers your questions | This page |
"Prompts won / lost", "Prompt coverage" and "Citation gatekeepers" carry a count once their data has loaded, for example "Prompt coverage · 48".
The window they share
Internal analysis has its own window selector with "7d", "30d" and "90d". It does not follow the client report's date range. The header reads, for example, "Period: last 7d (vs. prior 7d)": the window you pick is compared with the same number of days just before it.
- Briefing, Prompts won / lost, Prompt coverage and Citation gatekeepers open on 7d, and your choice travels with you when you switch tabs.
- Portfolio has its own selector with the same three windows plus "All-time", and opens on 30d. If you leave Portfolio on "All-time", the next tab reads 90d.
- Data Scientist has no window. It reads the whole project.
Prompt coverage
Prompt coverage is a table with one row per tracked prompt, covering the answers collected in the window: which prompts get answers at all, where you are named and where a competitor is named instead.
| Column | What it shows |
|---|---|
| Prompt | The prompt, with the AI engines that answered it listed beneath |
| Stage | The buying stage the AI assigned to the prompt. A dash means it is not classified yet |
| Responses | AI answers collected for this prompt in the window |
| Brand mentions | Answers that named your brand or domain. Each answer counts once |
| Mention rate | The share of this prompt's answers that named you. A dash means no answers were collected, not 0% |
| Competitor mentions | Answers that named at least one of your tracked competitors. Each answer counts once |
| Last seen | The last day an answer arrived for this prompt |
Answers that could not be linked to a tracked prompt are left out, so "Responses" can read lower than the won and lost view. Before any answers arrive, the tab reads "No prompt coverage data yet."
Portfolio
Portfolio is a board of tiles, titled "Analysis", about the opportunities first seen in the window (or across all time). Each data update refreshes it.
The top row
- Opportunities: how many pages are in the window.
- Actionable gaps: how many of them have been given a content-gap route.
- Detail generated: how many actionable gaps have their detail written, as a count and a share.
- Pitched: how many have a recorded pitch date, with how many are agreed and how many are live. Only a verified placement counts as live, so a yes is never reported as a win.
How the AI read the window
- AI buckets: how the AI grouped the window's pages by type, with a share beside each group.
- Priority score: how the window's priorities spread, with a summary beneath: Average, Median, P75, P95 and a count of high-priority pages. The priority is a 0 to 100 priority for outreach; the queue is ordered by it. One page's reasons are shown on the deck's "Promote?" card under "Why this score" and in its detail panel under the "Priority score" hint; see The detail panel.
- Outreach types: the outreach types the AI suggested most in the window, up to eight.
- Approach categories: the approach the AI suggested for the window's pages, with a share beside each.
Content gap themes
This section fills once some actionable gaps have their detail generated. Until then it says what to do next and links to Content gaps; see Content gaps.
- Top source weaknesses, Top missing sections and Top data opportunities: the themes the AI named most often across the gap analyses, each with a count.
- Recommended asset formats: the formats the AI recommended for your gaps, with a share beside each.
- Recent next actions: a sample of the next actions written for those gaps. Each one opens its opportunity in Content gaps.
Pipeline and outreach
- Status pipeline: the window's opportunities by their current status, each counted once, from "New" to "Live placement" with the closed statuses last. See Statuses.
- Outreach state: how far the window's opportunities have got, one row each: "With contacts", "With drafts", "Initial sent", "Pitched", "Agreed to place", "Awaiting verification" and "Live placements (wins)". Only verified live placements are wins.
- Follow-up backlog: scheduled follow-ups by when they fall due: "Overdue", "Due in 7 days", "Due in 30 days" and "Later" (more than 30 days out). Each tile opens the follow-ups due on Outcomes.
Data Scientist
Data Scientist is an AI analyst that explores your whole project and reports the patterns worth acting on. It has two views: "Analysis" and "Ask the data".
Generate an analysis
- Open Reports, press "Internal analysis", then "Data Scientist".
- Press "Generate analysis". The run shows in the engine bar, and its numbered steps appear live under "Analyzing your data…".
- Read the result: a headline, a summary, a row of key figures, then numbered findings. Each finding carries the analyst's stated confidence (High, Medium or Low) and four parts: "What we found", "The numbers", "What it means" and, where there is one, "Why it matters". Caveats follow, then "Show analysis steps", each step with a "Technical details" disclosure.
- Press "Regenerate" for a fresh read. The previous analysis stays on screen until the new one is ready.
A notice tells you when your data has changed since the analysis ran. A run longer than ten minutes stops waiting on the page: "Analysis is taking too long. Check back shortly."
Ask the data
"Ask the data" is a thread where you put a question to the analyst in your own words. Press a suggested question, such as "Which topics need our attention?", or type your own under "Your question", then press "Ask". Each answer is a written reply, finding cards where there are any, and the same "Show analysis steps" disclosure.
The thread belongs to the project, so your teammates' questions and answers are in it too. It shows the newest 40 messages; "Load earlier questions" brings the 40 before them. If a question fails, "Edit and retry question" puts it back in the box.
Limits
- Each analysis and question is AI work on your account. While the account is held, a press is refused with a line that points you to Billing. See Trial and billing and Plans and limits.
- A project the platform team has paused refuses new AI work with "This project is paused by the platform team. Automated work is on hold."
- "Generate analysis" is disabled while the analysis you started is running.
Good to know
- The briefing has its own share link, public to anyone who holds it; see Sharing, printing and briefings. No other Internal analysis tab reaches a link or a PDF.
- Prompts won / lost is visibility movement, not outreach. A prompt can be newly won with no pitch behind it. What outreach earned you is on Outcomes; see Reports overview.