We Measured Our Own Brand: 20 Questions, 90 Answers, 11% AI Visibility

Anyone who writes about AI visibility should at some point show their own result. Ours is not flattering: when someone asks an AI for a tool like ours, SEOLOK is not recommended in any answer we measured. This post reports that measurement as it is: the questions we asked, the answers that came back, how we read the numbers and the work that follows from them.

Every figure comes from the "AI visibility" screen of the SEOLOK dashboard, read on 12 October 2026. Nothing here is sample or illustrative data. The questions were asked in Turkish; they are translated below.

How the measurement was set up

Most of the first ten questions were about classic rank tracking. Before the second round we added ten questions that stand for the three needs we want to win: a Turkish AI visibility tool, ChatGPT brand tracking and GEO reporting for agencies.

The questions and their results

"Visible" is the number of answers that named us or cited our site, out of the answers measured.

The dashboard assigns the kind itself: questions containing our name are "brand", questions looking for options are "customer", questions that want to learn something are "information".

The numbers

Three notes on reading them:

The overall rate misleads on its own. All of the 11% comes from two brand questions. We are there when someone who already knows our name asks; we are absent when someone who does not know us asks for a tool. The number that matters is zero in 78 answers.

There is not enough data for a platform difference. Google AI Mode shows 7% and the other two 13%, but the ranges (2–21% and 5–30%) overlap almost entirely. We draw no conclusion from that gap.

There is no "drop". Per round the rate was 17% on 9 October and 8% on 12 October. That is not a decline: the second round added ten new questions, and we appear in none of them. The dashboard does not call it a change either; it compares only question and platform pairs measured in both periods and says there are not enough answers for that yet. Note the month you add questions the same way in your own reports.

Mentioned is not the same as cited

Our name appeared in 10 answers; our site was cited in only 3, all for one question ("Is seolok trustworthy?", from Gemini and ChatGPT). For "What is seolok?" the name is there, but the answer describes us from other sources, not from our own site.

In practice that means we do not control the content of the answer even on a brand question. The job is a short, clear, current answer to "what is SEOLOK" on our own site, and correct information on the outside pages that describe us.

Who was recommended instead

The names in the answers we were absent from cluster by topic:

• Rank checking and tracking questions: Google Search Console, the free rank checkers of Ahrefs and Semrush, AccuRanker, Nightwatch.
• AI visibility questions: Ahrefs AI Visibility Checker, Otterly.AI, Peec AI, Profound, Semrush AI Visibility Checker.
• The agency reporting question: AgencyAnalytics, DashThis.

One limit: this list is built from ChatGPT's own brand list and the brands we track. Brands we do not track are not counted in Gemini and Google AI Mode answers; we saw those by reading the answers one by one.

"Can you suggest a Turkish AI visibility tracking tool?" is a good example. In the answers of 12 October, Google AI Mode put a local tool and Ahrefs's free checker first, and ChatGPT put another tool with a Turkish interface first. Both grounded their answers in list and comparison articles. SEOLOK offers this product and was in neither answer, because we were in none of those articles.

The sources behind the answers

The sites cited most often across the 90 answers:

The top of the table is two kinds of site: products with their own free tool or comparison page, and blogs that publish "best tools" lists. Social media was cited in 6% of the answers and forums in 1%; forums do not decide this topic.

The work we took from this measurement

1. Pages that answer the question directly. For four questions we were absent from, the dashboard's "What to do?" card said the same thing: we have no page that stands out in searches close to that question. Separate pages for a Turkish AI visibility tool, ChatGPT brand tracking and agency GEO reporting are the priority for that reason.
2. Comparison pages. The answers cite competitors' comparison and "alternatives" pages. We need honest, sourced comparisons of our own with tools such as Otterly.AI.
3. Being in the list articles. Introducing the product, with correct information, to the authors of the "best AI visibility tools" articles the answers rest on. Not paid placement; access so they can try it and write.
4. Making the free tool visible. For "a free AI visibility test" three other tools were suggested and ours was not. The tool's page and the content that explains it need to be stronger.
5. Keeping the same list. The list is fixed now: 20 questions, every Monday. The next meaningful reading is possible once the same 20 questions have been measured at least twice more.

What this measurement does not say

• Traffic or sales. We measured mentions and citations; visits and conversions come from other reports.
• What your own screen will show. These are the answers a logged-out user sees from Türkiye; yours may differ.
• A lasting state. Two rounds are not a trend. Each new question was measured with three answers only; the dashboard shows no per-question rate below six answers, and we give only the counts here.
• Perplexity and Claude. Not measured.

We will update this post as the same questions are measured again, whether the result is good or bad. To set up the same measurement for your own site, the method is in how to measure AI search visibility; to try a single question now there is the free AI visibility checker, and for weekly measurement AI visibility tracking.