What we claim, and what we don’t.

Three independent states

Finding

yes, partial, no, unknown or not applicable. Unknown is neutral and excluded from any performance tally.

Verification

unreviewed, verified, conflicting or expired. Manual submissions start unreviewed. Only a trusted server-side reviewer workflow can change this — never the browser.

Collection

succeeded, not found, access blocked, failed or not configured. A failed or blocked collection is recorded as unknown, never as “no”.

Every claim carries

The criterion; the exact subject (a named branch or the brand as a whole); the claim type — observed fact, business-published claim, customer sentiment or interpretation; the source URL; source date if known; collection timestamp; an evidence excerpt or structured fact; the verification reason; a freshness expiry; and sample limitations.

What “verified” means

The attached evidence supports the narrowly stated claim. It is not a guarantee of the underlying real-world truth.

Known limitations

• Website presence does not prove responsive design or that checkout works.
• Ads not observed does not mean the business runs no ads.
• Reviews do not verify cleanliness, food safety or allergen safety.
• Job adverts do not prove staff training.
• Delivery ETA is a platform estimate, not measured delivery performance.
• Rating scales are recorded with every rating; no platform scale is assumed.
• Search and AI visibility require location, language, date, query or model, and repeated samples.
• Search snippets are discovery leads, not evidence.

Scores

We do not produce an overall score from how much evidence has been collected. Evidence coverage is shown on its own. We do not claim complete source coverage or perfect accuracy.