Right now running on UK and Romania locations
The AI Visibility Index measures how often a brand is named and how often its pages are cited when people ask AI. It reads two engines, ChatGPT and Google AI Overviews, across the keywords a website actually competes for in Google, and weights every answer by real Google search volume.
It runs alongside the SEOmonitor platform, on its own URL: https://www.seomonitor.com/ai-visibility.
Your rank tracking campaigns are unaffected: the platform tracks the keywords you have chosen, daily and weekly; the index reads a whole market from our research database, snapshot by snapshot. Use the index to find where you stand against competitors site and category level, find what to track and then track it in the platform.
The first report needs no account. A free account raises the daily limit. All SEOmonitor paid plans unlock the full unrestricted index, competitor comparison and CSV exports.
How the index reads AI engines
Five steps, and every number traces back to a keyword people actually search and an answer you can read for yourself.
1. You enter a domain, a brand, a keyword or a page
A domain gives you the full picture. A brand name shows how one name shows up. A keyword shows the AI answer behind one intent. A URL shows what one page earns as a source. Pick your market first.
2. One question in AI equals one keyword in Google
Every way people phrase a search collapses into one intent. We translate each keyword in our database into the single question a real person would ask AI, carrying the full intent, and run it once. In Google, "paracetamol" means "what is it and where do I buy it". In ChatGPT that same intent has to be asked as a question, so that is what we ask.
3. We run it through ChatGPT and AI Overviews
We ask the way real users ask, then read the answer: which brands get named, which domains get cited, and in what order. The engines are read separately, so you always see where you stand on each.
We use the current default ChatGPT, the same one a logged out person gets, because results track the web content being retrieved rather than the model version.
4. We weight it by real demand
Topics with larger demand count more than rare ones. Every score is weighted by verified Google search volume. Prompt volume estimates are guesses, so we do not use them. Keywords under 50 monthly searches are not processed.
5. You get the report
The verdict, the competitive map, the category breakdown and the keyword by keyword read: who AI recommends instead of you, and where the gap is.
What the metrics mean
Two metrics carry the report, mentions and citations - and both are important. A brand can be named without being cited, and a page can be cited in an answer that never names the brand.
% mentioned is the headline. The share of AI answers where the brand is named, weighted by search volume. Keywords where the engine names no brand at all are excluded from the calculation, so the score reflects only brand-competitive intents. Only real competing brands count: an answer that names youtube or a government portal is not a mention anyone can win.
% cited is secondary. The share of keywords where the domain's own pages are used as a source. Computed only where the engine shows citations: AI Overviews when one appears, ChatGPT in search mode.
Citation rank is per keyword: where your domain lands among that answer's cited sources. #1 means cited first. A dash means not cited for that keyword.
Per engine state is reported separately for ChatGPT and AI Overviews: mentioned, not mentioned, or no brands named. AI Overviews adds "not present" for searches where Google showed no overview at all, which is a different thing from not being mentioned.
Every percentage shows the underlying calculation. On screen, or on hover.
The four report types
Report | What it reads |
Domain | All of the domain's brands aggregated, plus everything its pages earn as citations, across both engines. The full report. |
Brand | Filtered to answers that name that brand specifically, and to the brand-competitive market only: the keywords where at least one real brand is named. Citations are associated with pages, so they stay in the domain report. |
Keyword | One search, one prompt, both engines: who gets named, who gets cited, and the full answer text. |
URL | The answers citing one exact page as a source, and the searches where that page could be cited but is not yet. |
A domain can own several brands and spellings, for example Procter & Gamble and P&G The domain view aggregates them all. Associations are detected automatically and can contain errors, so you can edit the brand list for your own reports.
Inside a domain report
Market context
Built on the keywords where this domain ranks in Google's top results, each carrying real monthly search volume. The set is curated. No irrelevant terms, duplicates, close variants folded into one so volume is never double counted, and no branded or navigational searches, because a brand always wins its own name and there is no competitive insight in it.
Under it, one card per engine:
ChatGPT: names a brand. The searches where ChatGPT names at least one real brand. This is the base for ChatGPT's % mentioned.
AI Overviews: AIO appears, then names a brand. An AI Overview can only name or cite where one appears, so the second bar reads inside the first.
Overall performance
One card per engine with the headline % mentioned and % cited, and sentiment inside the card on paid plans. Read the engines separately: they are separate reads, not two halves of one total.
Competitive map
Competitors are the brands competing for the same searches you do. If you compete for the same keywords, you compete for the same AI answers. Each row carries two bars: % mentioned on top, % cited underneath.
Switch the map between ChatGPT and AI Overviews.
Filter to a single category.
Add other competitors puts a brand of your choice on the map.
Performance by category
Every keyword in the database is classified once into product categories. The table shows, per category, the keyword count, the monthly volume, and % mentioned and % cited per engine, volume weighted.
Subcategories nest under their main category. Click the caret to unfold them in place. Same columns, same rules: each subcategory percentage divides by the domain's keywords in that subcategory.
Clicking a row opens the keyword table filtered to that category, scoped to the domain's ranking keywords. A green chip states the scope, and that same set is the denominator in the category table, so the two views reconcile.
View entire category switches to the whole category as it exists in the index, including searches this domain does not rank on. The chip turns blue, the keyword count and volume change with it, and the percentages read lower because they now divide by the whole category.
Sort by any column, or filter categories by name.
Comparison
The + Compare button on the category table and the keyword table adds a second website and reads it on the same keyword universe as this report. Same keywords, same denominators.
The default is the best performer on this report's keyword universe. Change it whenever you like to the domain you want to compare yourself against.
With a comparison running, each engine shows one metric for both websites, and a Mentions / Citations switch appears next to the control. Removing the comparison restores mentioned and cited in parallel.
Search landscape
The keyword table: every search where the brand is mentioned, cited, or could be because it competes for it. Ordered by keyword value by default, so the most relatable searches show first.
Per keyword you get the monthly volume, the mention state per engine, and the citation rank per engine. Controls above the table:
Filter by keyword text or by category.
Add a competitor you want to compare against
Sort by keyword value, search volume or citation rank.
Filter by state: All, Mentioned, Not mentioned, No brands, Cited, Positive, Negative. Set the engine to Any, ChatGPT or AI Overviews first.
Click a row to read the exact prompt and the exact answer.
No brands means AI named no brand at all for that search and gave generic advice. Those keywords stay in the table and can still carry a citation, but they are excluded from % mentioned.
Top cited pages
The pages of this domain that AI answers use as sources, ranked by the Google demand of the keywords they are cited on. These are the pages earning the domain's AI presence. Switch between ChatGPT and AI Overviews. Everyone sees the top five; the full list is paid.
The keyword and URL reports
Keyword report
How we read this intent: the keyword, and the single prompt it translates into.
What AI answered: the full answer per engine, with every brand named and every source cited. Switch engines to compare what each says to the same demand.
Extended topic: searches with a similar intent, not the same one. Each runs as its own prompt, and visibility rolls up across the topic. Switch between Mentioned and Cited, and between engines.
Share of voice across the extended topic: who owns the conversation, volume weighted. Paid.
URL report
Keywords citing this page: where AI answers cite this exact page today, with the demand behind them and the average position among the answer's citations.
Where this page can be cited: the searches where it could realistically be cited based on where it competes in Google, but is not yet, and the demand it is missing.
If a page has no citations yet, the report says so rather than showing an empty table.
Exports
Every table exports to CSV from the Export button in its section header. Export is a paid feature, and each plan has a row cap per table.
Report | Exportable tables |
Domain and brand | Performance by category, Search landscape, Top cited pages |
Keyword | Extended topic, Share of voice |
URL | Keywords citing this page, Where this page can be cited |
The export carries whatever the table is showing, including the active category filter, the state filter and the comparison columns. Filter first, then export, and the CSV arrives to your email.
Workflows
Find the topics AI recommends someone else for
Open your domain report and go to Performance by category.
Find a category where % mentioned is low against the volume it carries, and click into it. The keyword table opens filtered to that category and scoped to your ranking keywords.
Set the engine, then filter to Not mentioned.
Sort by search volume and read the answers on the biggest ones to see who is being recommended instead.
Export the list and add those keywords to your SEOmonitor campaign so you can watch them weekly rather than waiting for the next snapshot.
Compare yourself against one competitor on the same searches
On the category table, click + Compare and enter the competitor's domain.
Keep the metric on Mentions and read down the categories: the gaps are the categories where they are named and you are not.
Open the widest gap. The comparison carries into the keyword table, so you now have the searches behind it.
Switch to Citations to see whether they are also being used as a source, which tells you whether this is a content problem or an authority problem.
Export and work the list.
Research a competitor's whole market
Enter the competitor's domain instead of your own. You get their report: their market, their categories, their keywords, their competitive map. It is the fastest way to see which categories they are winning that you have not entered yet.
Research a category rather than a website
Search a category name to see the market itself: how many keywords it holds, how much demand, and which brands lead it. Open any of those brands from there.
Take an index finding into tracking
The index is a quarterly research snapshot, so it tells you where to look, not what changed this week. Once you have decided a set of keywords matters, add them to a SEOmonitor campaign. The platform then tracks AI Overviews daily and AI Search weekly, with the visibility trend explainer showing what moved and why.
Build a pitch for a prospect (agencies)
You do not need the prospect as a client, or an account of theirs, to read their AI visibility. The index reads any website in a live market from public data, so you can walk into a pitch with their numbers already in hand.
Run the domain report for the prospect and see where they stand right now in ChatGPT and AIO on all non-branded keywords they are ranking in Google
Open the competitive map and find where they sit. The gap between them and the leader can be a helpful angle to add in the pitch.
Add + Compare against the competitor they care about, on the category table. Read down the categories and pick the one where the gap is widest against real volume.
Click into that category and filter the keyword table to Not mentioned. Now you have the specific searches where AI recommends the competitor instead of them, with the monthly demand attached.
Open two or three of those keywords and read the actual answer. Showing a prospect the exact ChatGPT answer that names their competitor can be quite valuable.
Export the category table and the filtered keyword list as the appendix to your proposal to help scope the work from it.
Two notes for pitching. If the prospect ranks on very few keywords in Google, the report will say the market is small and offer to read them across a competitor's keywords instead; that borrowed read is often the more useful pitch, because it shows the market they are absent from. And because the index is a quarterly snapshot, position it as the diagnosis. The tracking of what your work then changes belongs in a SEOmonitor campaign, which is the natural second half of the proposal.
Coverage, cadence and plans
Engines: ChatGPT and Google AI Overviews.
Markets: UK and Romania are live. Every other market is listed and you can join its waitlist, which moves it up the queue.
Cadence: a single Q3 2026 snapshot today, refreshed quarterly for everyone.
Guests: one fresh report per type per day, no account needed. Anything you have already unlocked never counts again.
Free account: a higher daily report limit and the direct competitor filter, plus a wider data window on the keyword and category tables.
Paid: the full unrestricted index, sentiment, share of voice, your own competitors, comparison, the full cited pages list and CSV export.









