There is no official ranking for AI answers. So we do not pretend to give you a rank.
Keep answer samples, citations and visits separate
Save the exact question, date, assistant and response from your repeatable test set.
Where available, inspect the AI reports in webmaster tools. A citation count is not a ranking.
Review analytics for visits and enquiries. A visit does not prove which answer prompted it.
Bing's AI Performance report describes cited pages and grounding-query trends, while explicitly saying it does not measure rank or authority. Bing Webmaster Tools ↗
Write down what moved and what did not
| Signal | Month 1 | Month 2 | Careful reading |
|---|---|---|---|
| Named in fixed question set | 3 of 10 | 5 of 10 | More mentions in this small sample |
| Wrong location claims | 2 | 0 | A known fact error stopped appearing in these runs |
| Enquiries from AI referrals | 0 | 1 | Too little data to infer a trend |
OpenAI documents a utm_source=chatgpt.com parameter on links from ChatGPT Search, which can help identify some inbound visits. Google provides Search Console reporting for AI features. Neither source exposes every recommendation someone sees. OpenAI publisher guidance ↗ Google Search Console update ↗
Choose the next fix from evidence
If several saved answers repeat the same wrong location, check the site and public profiles. If a useful page gets cited but visitors do not enquire, examine the page's clarity and contact path. If nothing changed, report that plainly and keep the question set stable for another period.