Method
How Is AI Visibility Measured?
No engine publishes a complete rank or share metric. So we combine a prompt-based method with the native data engines do provide.
AI Visibility Turkey is the Turkish-market home of Zen Media, Optimum7 and ecommerce.com.
Method: run buyer questions and record the answers
This is the standard approach: build a prompt set from real buyer questions, run it across several engines and record the answers systematically.
Build the prompt set
For each product or category, buyer questions are written in Turkish and English. Buyers in Turkey ask in Turkish; buyers abroad often ask in English, sometimes in German, Arabic or Russian.
Run across several engines
The same questions are run on assistants such as ChatGPT, Gemini and AI Mode, Perplexity, Claude, Copilot and Grok.
Record the answer
Was the brand mentioned, was it cited, how was it described, which competitors and which sources appeared: all of it is recorded.
Segment
Prompts are grouped by persona (buyer, investor, patient, importer, traveler, procurement manager) and by intent (high-intent buying, comparison, reputation, problem-solving, risk validation).
Example prompts
The questions below are illustrative; they are not the result of a real measurement. A branded question (with your name in it) is for comparison only; the real goal is non-branded questions.
Which hair transplant clinic in Turkey is the safest?
Illustrative: not a measured result.
Best textile manufacturer in Turkey for private label
Illustrative: not a measured result.
İstanbul'da yatırım için en güvenilir konut projeleri hangileri?
English: Which are the most reliable housing projects for investment in Istanbul?
Illustrative: not a measured result.
Örnek Hukuk Bürosu iyi mi?
Illustrative: not a measured result.
Six metrics
This is a measurement framework we define ourselves; it is method, not statistics.
AI share of voice
In how many tracked prompts your brand is mentioned compared with named competitors.
Citation count and cited URLs
How often your domain is linked as a source and which pages are cited. Bing AI Performance provides this natively.
Mention accuracy and sentiment
How the brand is described and which factual errors need fixing.
Source mix
Which third-party domains engines cite for your category.
Crawl access
Successful hits from AI bots in server logs; pages indexed in Google and Bing.
Business outcome
Sessions and conversions from AI referrals. Google says visits from AI features show up in Search Console and recommends measuring the conversion value of visits.
Source (for the Google part): Google Search Central: AI features ↗; Google: Succeeding in AI search ↗
Native data sources and their limits
The data engines provide is valuable but partial. You also need to know what each one does not show.
Scroll the table sideways.
| Source | What it shows | What it does not show |
|---|---|---|
| Bing Webmaster Tools: AI Performance (public preview) | Total citations, average cited pages, grounding queries (sample), page-level citation activity and visibility trends across Copilot, Bing AI summaries and partner integrations | Placement within the answer; only citation counts are reported. It does not cover other engines (ChatGPT, Gemini, Perplexity). |
| Google Search Console: Performance, "Web" search type | Traffic from AI Overviews and AI Mode is counted inside these numbers | There is no AI-only report; AI traffic cannot be separated from classic search |
| Server logs | Successful requests from AI bots to your site; access blocks | Whether a bot actually used the page in an answer |
Source: Bing Webmaster Blog: AI Performance ↗; Google Search Central: AI features ↗
Why one question is never enough
AI answers change with every run and every engine.
Ask the same question twice and you may get a different answer; engines also behave differently from one another. So we do not treat a single answer as evidence. We run many prompts, across several engines and more than once, and report directional trends. If someone tells you "we asked ChatGPT today and it recommended you", that is an example, not a measurement.
Our measurement tooling as a group
ZAVI ("Zen AI Visibility Engine") is our group’s prompt-tracking platform: per-engine mention, citation and "Answer Share" scoring across Claude, ChatGPT, Gemini, Perplexity and Grok. We make no price or product promise here.
What a report contains
The report’s story answers four questions: who is winning, who is invisible, why AI recommends certain companies and what can be fixed.
Contents of the monthly report
- Executive summary
- Movement in AI answer share
- Model-by-model visibility changes
- Top winning and top missing prompts
- Competitor movement
- Changes in citation sources
- Movement in reputation sentiment
- Content and PR assets published
- Technical improvements completed
- Next month’s action plan
The layout on the left is an example; it contains no real client result. For reporting details see monitoring and the monthly program.
Limits
An honest measurement also says where it ends.
What we do not know and cannot measure
- No engine publishes a complete rank or share metric; our figures are directional trends based on sampling.
- Bing reports citation counts only, not placement within the answer.
- Google offers no AI-only report; the traffic sits inside the "Web" type in Search Console.
- We found no independent study measuring citation behavior for Turkish-language queries.
You can find the engines' own documents in the engine-by-engine guide, short definitions in the glossary and the order of work on the process page.
Let us measure where your brand stands today
Send us your website and topics and we will build the prompt set together. We reply by email.