What does AI say when buyers ask about your market?
A growing share of buying questions goes to ChatGPT first. AI visibility monitoring runs your buyers' questions against the major AI engines on a schedule and tells you whether you're mentioned, cited, and recommended - or losing invisibly to a competitor.
Set it up once. It watches forever.
Viz learns your brand
A discovery scan profiles your site, checks how you show up in search, and mines your Search Console queries (if connected) to understand what you sell and who you compete with.
You pick the prompts
Viz suggests the questions your buyers actually ask AI assistants - "best X for Y" - and you choose which to track. Add your own, and the competitors you want benchmarked.
AI engines get asked, on schedule
Your prompts run against multiple AI engines daily or weekly. Every response is stored, scored, and trended - no manual checking, ever.
Nine numbers that describe your AI presence.
Mention rate
How often AI answers name your brand at all - the gate every other metric sits behind.
Citation rate & rank
When answers cite sources, is your site one of them - and how high in the list?
Sentiment
Viz scores the tone of what AI says about you, so you catch a narrative going wrong early.
Share of voice
Head-to-head against the competitors you choose: who gets mentioned more, cited more, and recommended first for the prompts that matter.
Top cited sources
A leaderboard of the sites AI engines cite for your prompts. See exactly which content wins citations - including when it isn't yours.
Brand leaderboard
Every brand AI mentions across your tracked prompts, ranked. Discover competitors you didn't know you had.
First-paragraph mentions
AI answers front-load their verdict. Viz tracks how often the opening paragraph names your brand - the recommendation slot - not just a mention buried somewhere in the text.
Every metric, per model
Mention rate, first-paragraph rate, citation rate, and citation rank, each charted model by model - so you can see which engines already recommend you and which still need convincing.
The competitor matrix
Mention and citation rates for every tracked brand on every model, in one table. When a competitor owns one engine and you own another, this is where it shows.
The engines your buyers actually use.
The free plan starts you with one tracked prompt running weekly on Perplexity Sonar. Lite tracks 10 prompts daily across four engines - Gemini Flash, Claude Sonnet, GPT, and Sonar. Pro doubles the prompts and adds the frontier tier: Gemini Pro, Claude Opus, GPT Pro, and Sonar Pro. All queries run in the Viz cloud - your WordPress site never talks to an AI provider or holds an API key.
AI answers don't have a page two.
A search results page distributes attention: ten links, ads, maps, a chance for everyone. An AI answer concentrates it: two or three names, stated with confidence, often with no visible alternative. When buying research moves from the first format to the second, the market doesn't shrink - but the number of winners per question does. Brands that were comfortably discoverable on page one can become structurally invisible in answers, and nothing in their existing reporting will say so.
The inputs to those answers are knowable. Engines synthesize from what models learned in training and what retrieval fetches at question time - which is why your crawl data and your answer tracking belong in the same tool. If retrieval bots never fetch your comparison pages, your absence from answers is a reachability problem. If they fetch constantly and you still aren't cited, it's a content authority problem. Those two diagnoses have opposite fixes, and only seeing both sides tells you which one you have. Start with the crawl side in Analytics, or read how to measure AI referral traffic.
How to read a mention-rate chart without fooling yourself.
Single answers are anecdotes
Generative engines are probabilistic: the same prompt can name you today and skip you tomorrow without anything changing. One screenshot proves nothing in either direction. Rates over weeks are the smallest unit of truth in this discipline - which is why Viz reports trends, not transcripts.
Movement needs a suspect
When mention rate genuinely moves, look for the cause in the inputs you can observe: a competitor shipping content that now wins the citation, retrieval bots starting or stopping fetches of your key pages, or an engine swap on the provider's side. The crawl log plus the top-cited-sources list usually names the suspect.
Citations beat mentions
If you must pick one number to improve, pick citation rate on the prompts closest to purchase intent - it correlates with the visits and conversions Viz credits per assistant (GA4 adds session depth), which closes the loop between what AI says and what it sends.
Visibility questions, answered plainly.
How is AI visibility different from SEO rank tracking?
A rank tracker measures your position in a list of links; everyone on page one gets seen. An AI answer is a synthesis: it names a handful of brands and moves on, and there is no page two. AI visibility tracking measures whether you are in that synthesis at all - mentioned, cited, recommended - which is a different and harsher question than where you rank.
Why do answers change between runs of the same prompt?
Because generative engines are probabilistic and their retrieval sources update constantly. Any single answer is an anecdote. That is why Viz runs your prompts on a schedule and reports rates and trends - mention rate over weeks, not the wording of one response - which is the level at which the numbers become stable enough to act on.
Which prompts should I track?
The questions a buyer asks when they don't yet know you exist: "best X for Y" comparisons, problem-phrased questions, and alternatives-to-competitor prompts. Viz suggests candidates from your site profile and Search Console queries, but you choose the list - you know how your customers talk.
Does improving AI visibility mean gaming the engines?
No - and attempts to game them tend to age badly. AI engines synthesize from crawlable, citable, specific content. The reliable path is the honest one: publish authoritative pages on the questions you want to win, make sure retrieval bots can reach them, and measure whether the engines start agreeing. The measurement tells you if the content is working; the content does the work.
Where do the queries run?
In the Viz cloud, on a schedule. Your WordPress site never calls an AI provider and never holds an API key - the plugin stays lean, and the tracking works even if your site is behind a strict firewall.
How fast do content changes show up in AI answers?
It depends on which input you moved. Answers built on live retrieval can reflect a new or updated page within days of retrieval bots fetching it - watch your crawl log for the fetch, then your mention rate for the effect. Knowledge baked into model training moves on the timescale of model releases, months not days. This is why tracking trends beats checking answers by hand: the lag itself is information about which lever you are actually pulling.
Is AI visibility the same thing as AEO?
Mostly. Answer engine optimization (AEO) and generative engine optimization (GEO) are the market's names for the work of showing up in AI answers. Viz measures that work as AI visibility and, unlike a visibility-only tool, follows it through to the visits and conversions it produces.