AI visibility is how often and how prominently your brand is named in answers generated by AI models such as ChatGPT, Gemini, Perplexity and Google AI Overviews. It is measured as the share of tracked prompts in which your brand appears, weighted by its position in the answer.
A search engine gives a buyer ten links and lets them choose. An AI model gives them an answer, and that answer names two or three brands. Everything else is not ranked lower, it is simply absent. That is the shift AI visibility measures: not where you sit in a list, but whether you are in the room when the recommendation is made.
It is not the same as brand monitoring. Monitoring tells you that your name came up. AI visibility tells you how often it came up across the questions your buyers actually ask, in what order relative to competitors, in what tone, and which sources the model used to decide. Those four things are what you can act on.
Five metrics, tracked per prompt, per model and per market. A number without those three dimensions attached is not comparable to anything. Select a metric to see what it looks like in practice.
Illustrative example: Visibility
78%Named in 78 of 100 runs of a category prompt
| Metric | What it measures | What good looks like |
|---|---|---|
| Visibility | The share of tracked prompts where an AI model names your brand at all. | Above 30% in your core category |
| Position | Where your brand sits in the answer when it is named. A first mention carries far more weight than a fifth. | Average position 3 or better |
| Share of voice | Your mentions as a proportion of all brand mentions across the same prompts. | Ahead of your two closest competitors |
| Sentiment | How the model describes you when it names you, scored per answer rather than per prompt. | Above 55 and stable |
| Citations | Which URLs the model pulled from to build the answer, yours and everyone else's. | Your own domain in the top five cited sources |
Pick a question a buyer would ask an AI model and see which brands get named, and how often. This is the same view KIME builds for your own category.
How do you check your AI visibility?
Asking ChatGPT about your own brand once tells you nothing reliable, because the answer changes between runs. Checking it properly takes five steps.
Write the prompts your buyers actually type
Not keywords. Full questions, in the words a buyer would use, including the ones that never mention your brand. Twenty to fifty prompts is enough to start.
Set the markets and models that matter
An answer in ChatGPT in the US and an answer in Gemini in Denmark are different answers. Track each combination you sell into separately.
Run each prompt repeatedly, not once
Models are non-deterministic. A single run tells you almost nothing. Repeated runs over time turn a snapshot into a rate.
Read the citations, not just the score
The list of sources an answer was built from tells you exactly which pages you need to be on, and which of your own pages are already working.
Fix the gap where the citation is missing
If a competitor is named and you are not, the difference is almost always a source the model trusts and cannot find for you. That is where the work goes.
The two overlap, and pages that rank well in Google are still more likely to be cited by an AI model. But the object being optimised is different, and so is the scoreboard.
| Dimension | Traditional SEO | AI visibility |
|---|---|---|
| What ranks | A list of ten links, ordered by relevance and authority | A single generated answer that names a handful of brands |
| What you win | A click | A mention, and with it a place on the buyer's shortlist |
| The unit of measurement | Keyword and position | Prompt, mention, position within the answer, and sentiment |
| What moves it | On-page relevance, internal linking, backlinks | Earned media, third-party consensus, structured and quotable content |
| How fast it changes | Positions are relatively stable week to week | Answers are regenerated per query and can shift daily |
| Who you compete with | Whoever ranks for the keyword | Whoever the model considers a credible option, including brands that do not rank |
The practical consequence is that a page can rank first in Google and still be invisible in ChatGPT, because the model built its answer from a review site, a press release and a competitor comparison rather than from the brand's own landing page.
AI visibility services cover the measurement and the work that follows it. In practice they break into four parts.
- Monitoring, which measures where a brand currently stands across models, markets and prompts.
- Diagnosis, which identifies the specific sources and gaps causing the brand to be left out of an answer.
- Content and technical work, which makes a site readable and quotable for AI crawlers: structured data, clean server-rendered HTML, llms.txt, and pages that answer questions directly.
- Earned media and third-party presence, which is where most of the weight sits, because models lean on what other sites say about a brand rather than what the brand says about itself.
Most tools stop after the first two. KIME measures visibility, placement, share of voice, sentiment and citations across every major model and market, then turns the gaps into a prioritised list of actions, and with AWX can execute those actions rather than just recommending them.
We would love to have a chat
Book a demo tailored to your situation, and see how KIME can help you dominate your industry in AI search.