Read the visibility overview
Coverage, mentions, position, sentiment and trends, and how the time-range and engine filters work.
In shortThe Visibility overview summarises your stored prompt runs for one domain: Market Share in AI, AI Recommendation Rate, Average Rank, AI Citing Links and Reputation Score, a per-engine breakdown and a trend chart. It uses your latest result for each prompt on each engine within the chosen window. Filter by 7, 14, 30 or 90 days, engine or prompt. Daily trend points are grouped by UTC calendar day.
What the overview is for
Once you've run a handful of prompts, single results get hard to follow. The overview answers three questions at a glance: how often AI recommends you, how high you appear, and whether that's improving. Use it for a weekly check-in and as the source for client updates.
If you haven't run prompts in the selected window, the cards are empty or show older readings. Nothing refreshes on its own. Clicking Refresh reloads stored data; it doesn't run new prompts.
Open and filter it
- Open Visibility overviewGo to AI search → Brand visibility → Visibility overview and check the domain in the domain selector at the top.
- Pick a windowChoose 7d, 14d, 30d or 90d. These count back from right now, so 7d means the last seven times 24 hours.
- Narrow by engine or promptUse the engine dropdown to see one engine, or the All Prompts dropdown to focus on one question. A Filtered by prompt bar appears; click Clear filter or Reset to go back.
- Export if you need the dataExport downloads a CSV of the headline numbers for the current filters, such as share of voice, recommendation rate, average position and mentions. It's handy for your own charts or a client deck.
The five KPI cards
| Card | How it's worked out | Labels you'll see |
|---|---|---|
| Market Share in AI | Your brand mentions as a share of all brand mentions (yours plus competitors') in the answers | 40%+ Category Leader, 20%+ Competitive, below that Growing |
| AI Recommendation Rate | Share of answers that mention you, using the latest answer per prompt per engine. Shows change against the previous period | 70%+ Top Pick, 45%+ Strong, 25%+ Growing, below 25% Low |
| Average Rank | Your average position in answers where you appear | #1–2 Top Choice, #3–4 Alternative, #5+ Low Rank, no mentions Not Cited |
| AI Citing Links | Total cited links in those answers, with your mention count underneath | No label |
| Reputation Score | Share of answers with a positive tone about you | 60%+ Strong, 30%+ Neutral, below that Review |
AI Citing Links counts every cited link in the answers, not only links to your site, even though its tooltip says otherwise. To see how often your own pages are cited, use Citations. Reputation Score divides positive answers by all answers, including ones that don't mention you, so a low recommendation rate pulls it down too.
The one-line AI Briefing above the cards shows "#1.0 avg rank" even when you were never mentioned; the Average Rank card, which then shows a dash and Not Cited, is the one to trust. The recommendation card's subtitle says "% of searches citing you", but it counts mentions, not citations. And the share chart further down counts only your top five competitors, so it can show a higher share than the Market Share in AI card, which counts all of them.
Under the cards, AI Engine Breakdown shows the same picture per engine. A big gap between engines is normal, because each searches and ranks sources differently. It tells you where to focus.
The tabs below
- Overview & Trends: the AI Recommendation Rate Over Time chart, plus Tested Buyer Queries & AI Ranking Status, a table of every prompt with its status and best rank. Click a row to inspect the answers.
- Competitor Intelligence: a short version of the Competitors page.
- Action Plan: Recommended Actions to Grow AI Visibility, a list of suggested tasks from your results. It's a quick list, separate from the per-prompt Action plan page.
- Audit Log & Runs: every run, newest first, with engine, prompt and outcome.
Dates and the trend chart
The trend chart puts each run on the UTC calendar day it happened, not your local day. If you're in New York and run prompts at 10 pm, they appear on the next day's point. The same applies to custom date ranges on the Citations page: a date you pick covers that whole day in UTC, from midnight to midnight.
Run the same set of prompts on the same engines each time. If you add twenty new prompts this month, your recommendation rate changes because the questions changed, not because AI changed its mind. Keep the wording fixed too, because editing a prompt's text can split its history.
Check yourself
0 of 3 answered.
Progress is saved in this browser.
Put the lesson to work
Paste your site and get a free SEO, AEO and GEO audit with the fixes, then come back to the next lesson.