A real check of our own site
Captured 2026-10-02T00:05:25.602Z to 2026-10-02T00:05:26.181Z · HTTP 200
We fetched Asked & Found’s public homepage without JavaScript, using the same fetcher as the free checker. This is a real observation of our own site, not fictional sample data. It is a website-checker demonstration, not a paid customer story or an AI recommendation audit.
What we found and corrected
The earlier checker treated missing listing links and rating markup as fixable deficiencies. Its result text made unsupported claims about what assistants must do. No assistant was queried, and neither a review requirement nor a listing-link requirement was established by the HTML.
Earlier rating observation: No rating markup. Your reviews may be on the page, but machines can’t read the score from it.
Corrected observation: No rating markup was detected. Reviews and ratings may be readable as text or available from other sources. Do not add markup merely to pass this check; self-serving LocalBusiness reviews are ineligible for Google review snippets.
We made rating markup, listing links, structured data and business files optional observations, excluded them from the checker count, and removed the unsupported review and listing recommendations. The same recorded responses were passed through the revised analyzer.
What changed: Link to your other listings; Label the reviews on your page were removed from suggested work. This corrects our product’s advice. It does not show a visibility gain, better rankings, more leads or changed assistant answers. Old and new counts have different denominators and must not be presented as a visibility improvement.
Inspect the evidence
- Raw HTTP responses and original checker output — homepage text, inspected files, status, final URL and capture times. Baseline source commit: 7edf764851d9d4ace214694e889ff8b7157a3596.
- Revised analysis of the saved snapshot — every observation, suggested action and analyzer source hash.
- Comparison manifest and artifact hashes. Re-analyze the same recorded HTTP snapshot; not a new assistant run or network retest.
Limits and the next pilot
This was one public capture, with no account, private customer data or assistant API involved. Heuristic extraction can miss or misinterpret content; the output was reviewed against the saved responses. The tool does not render scripts or OCR images. An assistant could use other pages, listings, images or rendered content.
No real 20-question assistant report or paid pilot has been published. That requires a real business with permission, verified facts, exact questions, provider/model/interface and search settings, timestamps, raw answers, citations and human-reviewed corrections under the report protocol. The fictional report remains a separate format example.