The short version
Treat issuance, publication and observed display as separate findings. Public evidence can support a carefully scoped conclusion, but it rarely proves that every email stream or recipient sees a working logo.
Define the research question and domain
Choose what you want to establish: whether a mark certificate was observed, whether a particular domain publishes BIMI, or whether a logo appears in a specified inbox. Write that question before collecting evidence.
A company’s website domain may differ from its sending domains. If you have a legitimately received newsletter or receipt, note its visible From domain. Otherwise label the domain choice as an assumption. Avoid reporting a negative search on one guessed domain as proof that the company does not use BIMI.
Keep three evidence layers separate
| Observation | Careful conclusion |
|---|---|
| Certificate or precertificate in a monitored log | A certificate-related record was observed within that source’s coverage |
| Public BIMI record at a known name | That configuration was published when queried |
| Logo on a received message | A logo was visible in that provider, account and application context |
The BIMI Group’s issuer directory distinguishes certificate logs from revocation lists. A historical log entry is not, on its own, proof of a currently accepted certificate or an active deployment.
Make the findings reproducible
For each row, save the exact domain or query, source, timestamp, observed value and limitation. If you use the BIMI Data observatory, include its coverage and counting rules in the research notes.
Do not count renewals as new adopting organisations, or assume that multiple domain names represent independent businesses. Define the unit of analysis before calculating a total. When a source is unavailable or its coverage is incomplete, retain an “unknown” result instead of converting missing evidence into “no BIMI”.
Use evidence-sized language in the final report
“A record was observed for this domain on this date” is reproducible. “This company’s customers all see its logo” requires much more evidence. Separate what was directly observed from your interpretation of it.
For an inbox observation, also consider whether account or contact imagery could explain the picture. Keep personal message content out of shared research files. The aim is a defensible finding with clear limits, not the largest possible adoption count.
Sources & further reading
Provider guidance checked on . Requirements can change; check the original guidance when implementing. Planning checklists and examples are BIMI World’s editorial suggestions.