Method · Version 1.0 · November 2026
How do we measure?
Ölçtük is a count, not a list of opinions. We ask AI assistants the questions people ask, and count which apps appear in the answers.
1. Questions
Each category has three fixed questions, asked in Turkish. They stay the same every month so months can be compared.
- General: "Which is the best X app?"
- Turkey: "Which app do you recommend for X in Turkey?"
- Person: a specific need, for example "I'm a student, can you recommend a free X app?"
Some categories are also measured in English for comparison. The question text is published with each issue.
2. Engines and repeats
Questions are asked to ChatGPT, Gemini and Perplexity. Because AI answers change a little every time, each question is asked three times to each engine, each time in a fresh conversation. That makes 27 answers per category. The model name and whether web search was on are recorded with every issue.
3. Counting
- An app counts as "recommended" once for every answer in which it appears as a recommendation. Names mentioned only for comparison are not counted.
- Recommendation share = answers in which the app was recommended ÷ total answers.
- First place = number of answers in which the app was the first one recommended. It separates apps with equal shares.
- Versions of the same product (for example Lightroom and Lightroom Classic) count as one app.
4. Transparency
Every issue publishes all questions, the order of apps recommended in the answers, and the sources the AI assistants showed.
5. Independence and advertising
- Money never enters the ranking. No company can influence the measurement.
- Sponsored space sits outside the list and is always labeled "SPONSORED".
- The data is prepared by the Tocco Tech data team. If an app that is a Tocco Tech client appears in the index, it carries a "Tocco Tech client" note; Tocco Tech's own app carries a "Tocco Tech product" note.
6. App profiles
Every app that appears in a measurement gets a profile page (in Turkish). A profile contains only measurement results and public App Store facts. An app that wasn't measured can submit itself at submit@olctuk.com; every submitted app is measured within 30 days and gets a profile whatever the result. Profiles are free and their content isn't set by the app's owner; they can only ask for a factual error to be corrected.
7. Limits
This index measures what AI recommends; it doesn't say which app is better. AI answers can change with time, model and account. Pilot issues are small samples and should be read as indications, not conclusions.