Download the workbook
Use one row per run. Keep the exact question, product, date, visible search behavior, brand mention, cited URL, competitors, result state, and page version together.
Formatted for repeated manual tests. Includes dropdown fields, frozen headers, instructions, and example rows clearly labeled as examples.
What the log records
| Field group | What to capture | Why it matters |
|---|---|---|
| Test setup | Test ID, exact question, question type, product, date, and run number | Keeps repeated observations comparable. |
| Visible result | Visible web search, brand mentioned, cited domain, and cited URL | Separates what the interface showed from assumptions about hidden retrieval. |
| Context | Competitors surfaced, result state, page version, and notes | Connects observations to the page and change being tested. |
How to use it
- Write the questions before viewing any results.
- Include brand-free discovery, comparison, and known-brand questions.
- Run the same question more than once and give every run its own row.
- Copy cited domains and URLs exactly as shown.
- Label no-search or unavailable states separately.
- After changing the page, record the page version and repeat the same test set.
For the full procedure, read How to check whether AI engines cite your website. For the page review that comes before testing, use the GEO audit checklist.
What the workbook does not establish
The log records observations. It does not reveal hidden retrieval, prove why a source appeared, or establish a permanent ranking. Product behavior, location, timing, and other conditions may change the visible response. Keep those limits with any summary you publish from the log.
Audit the page before you retest it.
Enter the page URL to identify clarity, evidence, and quotability issues worth changing.
Audit a page free