What should buyers know about AI search visibility services?
What an AI search visibility service delivers, what a consultant does, who publishes the changes, what you provide and what nobody can promise.
An AI visibility score is the share of AI answers that name your brand. We ask ChatGPT, Claude, Gemini and Google AI Overviews the buyer questions you agreed with us, without your brand name in the question, and divide the answers that mention your brand by all valid answers. A brand named in 36 of 240 answers scores 15%.
Published 16 November 2026 · By Balázs Turán
Visibility = answers that mention your brand ÷ valid answers to questions without your brand name × 100, rounded to a whole number.
The steps:
You approve the buyer questions for each market language. A market language is one country in one language.
Each question goes to ChatGPT, Claude, Gemini and Google AI Overviews three times per market language, at least one hour apart, because AI answers vary from one request to the next.
An answer counts as a mention when it contains your brand name, a known alias or your domain, regardless of upper or lower case.
Failed requests, and searches where Google shows no AI Overview, leave the count. They are not counted as answers without your brand.
Mentions divided by valid answers, times 100.
| Case | Counts as a mention? |
|---|---|
| The answer names your brand or a known alias | Yes |
| The answer names your domain | Yes |
| Your page is only listed as a source, the text does not name you | No, it counts under "Own pages cited" |
| The request failed or Google showed no AI Overview | Not counted at all |
The question set excludes your brand name on purpose. A buyer who already types your name will usually find you. The score measures whether AI services name you when the buyer does not know you yet.
A brand measures 20 buyer questions in one market language, Germany in German.
| Step | Value |
|---|---|
| Questions | 20 |
| AI services | 4 (ChatGPT, Claude, Gemini, Google AI Overviews) |
| Answers per question and service | 3 |
| Planned answers | 20 × 4 × 3 = 240 |
| Valid answers | 240 |
| Answers that name the brand | 36 |
| Visibility score | 36 ÷ 240 × 100 = 15% |
If Google showed no AI Overview for 12 of its 60 searches, those 12 leave the count: 36 ÷ 228 × 100 = 15.8%, shown as 16%.
The readiness score is a weighted average of category scores, each from 0 to 100:
| Category | Weight |
|---|---|
| Content depth and quality | 30% |
| Entity clarity | 15% |
| Answer-first writing | 15% |
| Structure and extractability | 12% |
| Structured data | 10% |
| YouTube captions, only with a declared video channel | 8% |
| Meta and social tags | 4% |
| Sitemap | 3% |
| Hreflang, only for sites with several markets | 3% |
When a category does not apply, for example YouTube captions without a video channel, it leaves the calculation and the other weights are scaled up to 100%. llms.txt and AI crawler settings are checked and reported, but they do not change the score.
Example for one category, structure and extractability on 20 pages:
| Signal | Points |
|---|---|
| Headings: one H1 and at least two H2s | 30 of 30 |
| Lists on the page | 24 of 25 |
| Visible question-led answers on 60% of pages | 15 of 25 |
| Tables on 2 of 20 pages | 2 of 20 |
| Category score | 71 of 100 |
No. The readiness score measures your website. The visibility score measures AI answers. A site can score 90 on readiness and still be named in few answers, for example when the cited sources are other sites.
AI services give different answers to the same question. Three answers per question, AI service and market language, at least one hour apart, show whether a mention is repeated or a one-off.
ChatGPT, Gemini, Google AI Overviews and Claude are measured every month, with three answers per question on each. Each result keeps its question, market language and date.
No. A citation counts under "Own pages cited". The visibility score counts only answers whose text names your brand, an alias or your domain.
The Advisor shows the score, the answers behind it and the next actions for every question.
What an AI search visibility service delivers, what a consultant does, who publishes the changes, what you provide and what nobody can promise.
Which data an AI visibility measurement uses, where it is stored, the legal basis, which services process it and how to request a data processing agreement.
A worked example: five markets and three languages, how many market languages that makes, and the monthly price with Creative Data Engineers.
Five steps to improve answer engine optimization: measure AI answers, fix crawler access, write answer-first pages, work on the cited sources and measure again.