Documentation
Diagnosing why you lose
Who gets named instead, what the engine searched and read, and whether AI can read your site at all.
A number tells you where you stand. Diagnosis tells you which of several very different jobs to do next.
The competitive view
For every prompt you lose, see who the engine named instead and the exact pages, threads or listings it cited to do it. Aggregator hosts are never reported as competitors you could beat, and platforms like YouTube are labeled platforms rather than rival brands.
The retrieval funnel
Open a prompt and see the middle of the chain: the queries the engine issued while answering, the pages it read, which of those it cited, and whether you were named. One probe read 22 pages and cited 3. Your page can sit inside the 22 and never reach the 3, and that distinction is the whole work order.
| What the funnel shows | The job it implies |
|---|---|
| Not among the pages read | Show up for what the engine actually searches. A ranking job. |
| Read but not cited | Fix the page you already have. An answerability job. |
| Cited but not named | The engine used your page without knowing it is you. An identity job. |
The funnel runs on ChatGPT, the consumer app, because that is the surface whose payload exposes its retrieval step. A live Gemini probe carried neither the searched queries nor the results. Every other engine reads "not exposed", which means not exposed, never nothing considered.
Crawl truth
We knock on your site wearing each critical AI crawler's user agent and diff the result against a control fetch, which catches firewall and CDN blocks a robots.txt parser cannot see. The on-page audit then ranks retrievability findings by the Search Console traffic of the page each one sits on.