What is AI visibility?
AI visibility is how often — and how prominently — AI assistants like ChatGPT, Google AI Overviews, and Perplexity name your business when people ask what to buy or who to hire. It is measured per question and per market: your mention rate, your rank inside the answer, and your share of voice against competitors.
Customers increasingly start a purchase by asking an assistant, not a search box. The reply is not ten blue links — it is a short, confident shortlist: a handful of products or providers, each with a reason. That shortlist is the new shelf. If you are on it, you are in the running. If you are not, there is no page two.
The decision now starts before the click
Traditional analytics begins when someone lands on your site. But when an assistant answers “best wireless earbuds under €150” or “best electrician near me” directly, the consideration set is formed before any click happens — and often no click happens at all. Nothing shows up in your analytics, your ad platform, or your search rankings. A competitor can quietly own the answer for months while your dashboards look normal.
AI visibility is not a search ranking
A search ranking is a position in an ordered list of links, and it is relatively stable. An AI answer is a generated shortlist, and it behaves differently in three ways that surprise most teams:
- Borders change the shelf. The same question asked from Germany, the UK, and the US returns different brands, different retailers, and different sources. There is no single global answer — which also means a brand invisible in one market may already own another.
- Phrasing changes the shelf. In one of our scans, a sustainable audio brand appeared in 0% of answers for “best wireless earbuds” and 100% of answers for “best repairable earbuds” — same product, same market, same week. Visibility is a map across questions, not one score.
- Membership is steadier than order. Ask the same question repeatedly and the same brands tend to appear, but their order reshuffles. A single manual check can tell you rank #2 today and #5 tomorrow without anything real having changed — which is why one-off checks mislead.
The metrics that matter
| Metric | What it tells you |
|---|---|
| Mention rate | The share of answers that name you at all, per question and market. The foundational number — counted against answers actually given, not against every attempt. |
| Rank | Where you appear inside the answer when you do. First mention carries the recommendation weight. |
| Share of voice | Your mentions as a share of all brand mentions on the shelf — the competitive view. |
| Trigger rate | How often the engine answers the question at all. Some surfaces answer only a fraction of queries; reporting visibility without it flatters or hides the truth. |
| Sources | The publishers, review sites, retailers, and forum threads the answers cite. This is where visibility is actually won — see below. |
How to measure it credibly
Pasting a question into ChatGPT once is an anecdote, not a measurement. A credible read of your AI visibility needs five properties:
- Real buyer questions — the category queries customers ask (“best natural face cream for sensitive skin”), not your brand name.
- Each market separately — asked from the market's own context, because the shelf changes at every border.
- A logged-out baseline — no account history or personalization coloring the answer; measure what a new customer sees.
- Repeated sampling — enough answers per question to separate signal from shuffle.
- The full answer text, kept — so every number can be audited against what the assistant actually said.
What actually moves it
Answer engines assemble shortlists from what they know and what they retrieve — comparison articles, review sites, forum threads, retailer listings, and your own product data. Three levers follow:
- The sources. Find which publications and communities the answers in your category keep citing, and earn presence there. A review you're missing from is a shelf you're missing from.
- Your own data. Consistent specifications, clean structured data, and unambiguous category language make you easy to recommend. Conflicting specs across pages read as uncertainty.
- Positioning. Most brands already own a niche question the model associates with them. Finding it — and the high-volume questions where no one dominates yet — is the map for everything else. That improvement discipline has a name: answer engine optimization (AEO).
See your own shelf. A free snapshot scans one category across your markets and shows your mention rate, the competitors taking the shelf, and the clearest gap we find.One category · your markets · no credit card
Get your free snapshot →