Evidence seal
Evidence collection and scoring method
The method limits paid-ranking influence: discovery demand tells us what to investigate, but admissible evidence controls the signal.
AI assists evidence collection; it does not settle the facts
We use AI to find and organise evidence, then rule-based scoring reduces paid-ranking influence. Sources, dates, unresolved limits and a correction route stay visible. This process cannot guarantee complete or error-free information.
Evidence ladder
| Tier | Example | What it can support |
|---|---|---|
| Tier 1 | RDB, law, health authority | Regulator, law, official action or support route. |
| Tier 2 | Local or recognised independent publisher | Dated corroboration with attributable claims. |
| Tier 3 | Operator terms | What the company states about domain, entity and rules; not independent proof. |
| Tier 4 | User review or forum | Research context, not proof of legality or safety. |
| Discovery | Affgate or search demand | Research priority only. |
Signal rules
Green requires a current primary local record matching exact domain and operator. Red requires an official adverse record or corroborated documented evidence. Everything else is amber.
Corrections
A new source changes a signal only when it has a direct URL, publisher, date, jurisdiction and narrow supported claim. Submit a correction.