Practical buyer checks methodology

What this edition measures

A practical set of checks a buyer can understand before deciding whether to proceed.

Weights

Calculation

The observed signal score is 100 × detected weight ÷ assessed weight. Not-assessed signals do not enter that denominator. Coverage is the proportion of selected signals with a stored observation. Missing, blocked and legacy data remain explicit unknowns.

Evidence levels

At least 50% coverage is needed for a descriptive evidence level. Scores of 90+, 75+, 45+ and below 45 are labelled strong signals, generally positive signals, mixed evidence and limited evidence. A recorded threat-provider match takes precedence as elevated concerns. These labels describe selected observations, not a transaction recommendation.

Confidence and freshness

Confidence is medium when preserved automatic evidence has at least 70% coverage and is no older than 30 days; otherwise it is low. No report is labelled high confidence because the scanner cannot validate real-world fulfilment. The original scan date is preserved.

Publication and indexing

Reports can be read without being offered for indexing. Indexability requires preserved original observations, source URLs, at least 70% coverage, a scan no older than 30 days and a store classification. Incomplete laboratory and checkout profiles remain noindex until their distinct missing measurements are implemented. Source noindex and publication decisions are respected.

Limits

• Whether the seller will deliver an order cannot be established by a website scan.
• Available payment protections must be checked directly before payment.

Scores and reviews

The signal score is calculated from stored automated observations. Customer reviews are shared from Web Safety Index and displayed separately; their ratings do not enter this calculation.

Automated analysis does not guarantee the safety of a future transaction.