Method
The Kaynaq Method: Question → Source → Citation → Measurement
Four steps, repeated every month in the same order. Each step has a concrete output that becomes the next step’s input.
The Kaynaq Method has four steps: turn the questions buyers ask AI into a fixed question set (Question), prepare the site for AI crawlers (Source), get the brand mentioned accurately in trusted third-party sources (Citation), and ask the same questions every month on the same engines to report the result (Measurement).
1. Question: which answers should you appear in?
We start with questions, not keywords: what your sales and support teams hear most, Search Console queries, search autocomplete suggestions, and the answers where competitors show up.
- Output: a prioritised question set, with intent (information, comparison, purchase) and language for each question.
- Output: the list of engines to measure.
- Output: a baseline measurement: a dated record of where you stand today.
2. Source: is your site readable and trustworthy?
A page can only be quoted if it can first be reached and then understood. We set up crawler access (robots.txt, server-rendered text, speed), structured data (organisation, service, Q&A, article), consistent business details everywhere, and a short answer at the top of each page.
- Output: a prioritised technical task list and a log of changes made.
3. Citation: where does AI see you?
While measuring, we also record which sources each answer relies on. That shows which directories, lists, news sites and communities are cited for your questions, and we work to get your brand into them with accurate information.
- Output: a source map and a monthly citation work list. No paid links.
4. Measurement: what changed?
| Metric | Definition | How it is recorded |
|---|---|---|
| Mention | The brand is named in the answer text | Yes / no, per question and engine |
| Citation | The answer links to the brand’s site as a source | Yes / no and the linked page |
| Accuracy | What the answer says about the brand is true | Correct / partly / wrong, with a note |
| Competitors and sources | Other brands named and sites cited in the same answer | As a list |
Every question is asked twice on every engine and both runs are recorded, because even the same question can get a different answer. Date, language, location and account state go into the report. Results arrive as a monthly table, the change from last month and next month’s work list.
What this method does not guarantee
AI answers change with the model version, the user and time; no agency can guarantee a given position for a given question. We guarantee the work and the measurement: the same method, the same questions, a dated record, every month.
Frequently asked questions
01Does the question set change over time?
Core questions stay fixed so comparisons hold. New questions are added as a separate appendix and flagged in the report.
02Whose account is used for measurement?
To reduce personalisation we measure logged out or with clean profiles where possible, and record the conditions (date, location, language, account state) in the report.
How to cite this page
Kaynaq. (26 September 2026). The Kaynaq Method: Question → Source → Citation → Measurement. https://kaynaq.pages.dev/en/method
Updates and corrections
No corrections on this page yet. If you spot an error, write to us; we publish every fix here with its date.