What the platform measures, and what we can share

Two sources, both your own. Search Console and Google Analytics 4 supply clicks and sessions; your access log or CDN supplies the crawler hits themselves. Nothing is modelled or estimated from a third-party panel.

No code change. Cloudflare connects with one read-only API token, Vercel, Netlify and Fastly send their log drains, WordPress uses the Rixe plugin, and any other server can pipe its access log with a single command.

The AI crawlers that actually reached the page, matched by user agent — so you see which crawler hit which URL and when, not an estimate of how often you were probably crawled.

From the engines themselves, read on a schedule, plus the accounts you connect. Nothing in the panel is bought from a third-party panel or estimated from clickstream.

Daily on every plan. A market or engine you add starts on the next run rather than waiting for a billing cycle.

From the day the project is created. We do not backfill, because a backfilled answer would be a guess about what an engine said last quarter.

No. Every prompt in your set is asked on every run, on every engine you have on. A sample would hide exactly the rare answer that names a competitor.

The run records which model answered. When a version changes, the chart marks it, so a step in the line is explained rather than mysterious.

Yes — CSV from any panel, and the read-only Data API on Scale. The export carries the answer text, engine, market and timestamp.

For the life of the project, so a year-old answer is still openable. Deleting the account deletes them.