Authentication evidence pack for a deliverability incident
A practical guide to email authentication incident evidence: evidence, safe checks, edge cases, and a controlled next step.
Product & Engineering · September 25, 2026 · 2 min read
Email authentication incident evidence is best handled as a bounded evidence problem, not a guess based on one dashboard label. Immutable incident evidence is the useful starting point: collect the exact identity, record, header, or SMTP exchange involved, then compare it with what the receiver actually evaluated.
The short answer
Provide a neutral evidence template and chain-of-custody notes; do not infer root cause from one header alone. The safe workflow is to inventory the current state, test the real sending path, make one reversible change, and verify the result with fresh evidence. This page focuses on that task rather than repeating a general SPF, DKIM, or DMARC introduction.
Decision table
| Question | Evidence to collect | Decision boundary |
|---|---|---|
| Identity | immutable incident evidence | Use the exact domain, selector, host, or header field |
| Validation | capture DNS, headers, SMTP, reports, and change history | Record source, resolver/receiver, timestamp, and result |
| Edge case | one header is not enough to prove root cause | Do not infer a broader outcome than the evidence supports |
| Recovery | Saved prior state and named owner | Change or roll back deliberately |
Practical workflow
- Define the question in terms of immutable incident evidence. Write down the sending stream, domain or selector, provider, owner, and time window. If the question came from a dashboard, keep the raw message, DNS answer, SMTP reply, or report that supports it.
- Collect evidence before editing configuration: capture DNS, headers, SMTP, reports, and change history. Compare authoritative and recursive DNS where relevant, and compare a real received message with the configured values. A single local lookup or vendor status page is not a complete test.
- Make the smallest controlled change that addresses the observed failure. Preserve the prior value, approval, exact diff, and rollback owner. Keep SPF, DKIM, DMARC, DNS, transport, and placement observations in separate fields so one pass does not mask another failure.
- Re-test the same path and at least one representative edge case. Interpret the result conservatively: one header is not enough to prove root cause. Record what remains unknown, who owns the next check, and when the configuration should be reviewed again.
Edge cases and limits
The most important boundary is this: One header is not enough to prove root cause. Resolver caches, forwarding, multiple sending streams, provider-specific processing, and delayed messages can make two apparently similar tests disagree. Label an outcome as observed, inferred, or unknown. Do not turn an unmeasured demand signal or one successful test into a guarantee about delivery, placement, or receiver behavior.
RepMail relevance
For a RepMail sending stream, use the same evidence discipline before changing domain authentication: preserve the raw header or DNS answer, identify the stream owner, and keep a rollback record. The email authentication hub provides the broader map; the links below are the closest task-specific references.
Further reading
Read Authentication Results · Email Authentication Rollback Plan · Provider Specific Deliverability Triage
References
Resources in this guide
| Question | Evidence to collect | Decision boundary |
|---|---|---|
| Identity | immutable incident evidence | Use the exact domain, selector, host, or header field |
| Validation | capture DNS, headers, SMTP, reports, and change history | Record source, resolver/receiver, timestamp, and result |
| Edge case | one header is not enough to prove root cause | Do not infer a broader outcome than the evidence supports |
| Recovery | Saved prior state and named owner | Change or roll back deliberately |
Common mistakes to avoid
- Treating one dashboard label or local lookup as proof of end-to-end authentication.
- Changing several DNS or sender settings at once, which removes the comparison point.
- Assuming a pass for one identity proves alignment or placement for every identity.
Frequently asked questions
What should I check first for email authentication incident evidence?
Start with the exact identity and raw evidence: immutable incident evidence. Capture the source, timestamp, resolver or receiver, and current configuration before making a change.
Can one successful test prove the issue is fixed?
No. A test is time- and path-specific evidence. Repeat the relevant DNS, message, SMTP, or report check for each active stream and record its provider context.
What should I preserve before changing authentication settings?
Save the current DNS answer or message evidence, the owner and timestamp, the proposed diff, and the rollback decision. This makes before-and-after comparison possible when caches or delayed mail obscure the result.
Continue learning
Review the email authentication hub
Use the hub to choose the next protocol or diagnostic guide.