Reading DMARC aggregate reports and rolling out to p=reject safely
DMARC aggregate reports tell you, per sending source, how much mail passed or failed SPF and DKIM alignment — the only reliable way to know whether moving your policy to quarantine or reject will block legitimate mail or just spam. AdminKit's DMARC check and this guide walk through reading a report and a staged, safe rollout.
By AdminKit team, Netbase JSC · Updated
Key takeaways
- Publish
p=nonewith arua=address first. Aggregate reports then show, per sending IP, how much mail claiming your domain passed or failed DMARC, without changing delivery. - Fix every legitimate source that fails before tightening the policy, then step up:
none, thenquarantine, thenreject. - RFC 9989 (May 2026) replaced RFC 7489, removed the
pctpercentage tag, and addedt=ytest mode for trying a stricter policy one level softer first. - A
ruaaddress at another domain only works if that domain publishes a_report._dmarcauthorization record. - Keep reading reports after
p=reject: a new sender you forgot to authorize fails silently.
Why do DMARC aggregate reports exist?
They exist because a domain owner otherwise has no way to see who is sending mail as their domain. RFC 7489 (see the IETF Datatracker record) built reporting into DMARC from the start, and its May 2026 successors keep it: RFC 9989 defines the policy, and RFC 9990 defines aggregate reports.
The rua tag in a DMARC record names where to send a DMARC aggregate report (RUA): an XML summary of how much mail from each source passed or failed SPF and DKIM alignment, and what the receiver did with it. RFC 9989 says receivers should send them at least once every 24 hours, and RFC 9990 describes reporting periods that typically cover one UTC day. Without these reports, moving to a stricter policy is a guess.
ruf (failure, or forensic) reports are the other reporting mechanism DMARC defines. They describe individual failing messages, formatted as an ARF report, and are now specified in RFC 9991. That RFC notes they can carry header fields or whole messages, which may contain personal data. dmarc.org’s overview describes DMARC’s value as the feedback loop between senders and receivers, and that loop runs on aggregate data: Google, for example, states that Gmail doesn’t support ruf. Plan a rollout around aggregate reports; treat failure reports as a bonus.
How do you publish the rua address?
Add rua=mailto: to your DMARC record. If the address is at a different organizational domain, for example a DMARC reporting service, that domain must also publish an authorization record. An illustrative pair for example.com sending reports to example.net:
; DMARC record at the domain you protect, starting in monitoring mode
_dmarc.example.com. TXT "v=DMARC1; p=none; rua=mailto:dmarc-reports@example.com,mailto:rua@example.net"
; Authorization published by the report receiver's domain (RFC 7489 §7.1, RFC 9990 §4)
example.com._report._dmarc.example.net. TXT "v=DMARC1"
The check exists to stop abuse. Without it, anyone could publish a DMARC record pointing reports at a victim and flood them. A receiver that finds a rua address outside your organizational domain prepends _report._dmarc to the address’s host, prepends your domain, and looks up a TXT record there. If no record starting with v=DMARC1 comes back, the external address is ignored. A report receiver willing to accept reports for any domain can publish a wildcard such as *._report._dmarc.example.net.
Google’s report documentation adds a practical point: every mail server you send to may send its own daily report, so large organizations can receive hundreds or thousands a day. Google recommends a dedicated mailbox or group for them.
What is inside an aggregate report?
A report is an XML file, normally gzip-compressed, with three parts: report_metadata (who reported, and the date_range as Unix timestamps), policy_published (the DMARC record the receiver found), and one or more record elements. Each record has a row (source IP, message count, DMARC verdict), identifiers (the From domain) and auth_results (raw SPF and DKIM results).
The trimmed excerpt below is an illustrative example for example.com using documentation addresses, not a real report. Element names follow RFC 7489 Appendix C and RFC 9990 Appendix A, which share these core elements; RFC 9990 reports may also carry newer elements such as testing and np in policy_published, while RFC 7489 reports carry pct:
<feedback>
<report_metadata>
<org_name>receiver.example.org</org_name>
<email>noreply-dmarc@receiver.example.org</email>
<report_id>example-0001</report_id>
<date_range><begin>1794096000</begin><end>1794182400</end></date_range>
</report_metadata>
<policy_published>
<domain>example.com</domain>
<p>none</p>
<sp>none</sp>
</policy_published>
<record>
<row>
<source_ip>198.51.100.25</source_ip>
<count>1540</count>
<policy_evaluated>
<disposition>none</disposition>
<dkim>pass</dkim>
<spf>fail</spf>
</policy_evaluated>
</row>
<identifiers>
<header_from>example.com</header_from>
<envelope_from>bounces.esp.example.net</envelope_from>
</identifiers>
<auth_results>
<dkim><domain>example.com</domain><selector>esp1</selector><result>pass</result></dkim>
<spf><domain>bounces.esp.example.net</domain><scope>mfrom</scope><result>pass</result></spf>
</auth_results>
</record>
<record>
<row>
<source_ip>203.0.113.77</source_ip>
<count>212</count>
<policy_evaluated>
<disposition>none</disposition>
<dkim>fail</dkim>
<spf>fail</spf>
</policy_evaluated>
</row>
<identifiers>
<header_from>example.com</header_from>
<envelope_from>example.com</envelope_from>
</identifiers>
<auth_results>
<spf><domain>example.com</domain><scope>mfrom</scope><result>fail</result></spf>
</auth_results>
</record>
</feedback>
The first record says 1,540 messages from 198.51.100.25 passed DMARC through aligned DKIM. SPF passed in auth_results, but for the provider’s own bounce domain, so it failed alignment in policy_evaluated. The second record is 212 messages from an address that failed both checks with no DKIM signature at all. Because the policy is none, both were delivered (disposition none).
How do you read each row, and what should you do about it?
For each source IP, ask one question: is this one of my legitimate senders? Then match the row to a pattern:
| What the row shows | What it means | Action |
|---|---|---|
dkim pass, spf pass | Fully aligned legitimate mail | None |
dkim pass, spf fail, SPF pass in auth_results | A sender using its own bounce domain; DMARC still passes via DKIM | Usually none; DKIM carries it |
dkim fail, spf pass from a known sender | DMARC passes on SPF only, which forwarding can break | Set up aligned DKIM signing at that sender |
Both fail from a known service | A legitimate sender that would be blocked under enforcement | Add it to SPF or enable aligned DKIM before tightening |
Both fail from an unknown IP | Likely spoofing or an unlisted sender | Investigate; this is what quarantine and reject are for |
A reason element such as mailing_list or local_policy | The receiver overrode your policy | Review; no record change needed |
Reading raw XML across dozens of reports doesn’t scale. RFC 9989 recommends machine parsing, and Google recommends a third-party DMARC service. AdminKit’s DMARC report analyzer reads one report you paste or open, in your browser, and lists per source IP the volume, SPF and DKIM alignment and the applied policy. It does not identify who owns an IP, so match addresses against your own list of senders.
How do you roll out from p=none to p=reject without breaking mail?
Move one step at a time, and only after the reports show no unexplained failures from legitimate sources. The stages below follow RFC 9989 §5.1 and Google’s recommended rollout:
| Stage | Record (_dmarc.example.com) | What receivers do with failing mail | Move on when |
|---|---|---|---|
| 1. Monitor | v=DMARC1; p=none; rua=mailto:dmarc-reports@example.com | Deliver normally and report | All legitimate sources pass; at least one week, often longer |
| 2. Test quarantine | v=DMARC1; p=quarantine; t=y; rua=... | Apply one level lower (none) under RFC 9989 | Reports stay clean |
| 3. Quarantine | v=DMARC1; p=quarantine; rua=... | Treat as suspicious; Google sends it to spam | No legitimate mail in spam |
| 4. Test reject | v=DMARC1; p=reject; t=y; rua=... | Treat as quarantine under RFC 9989 | Reports stay clean |
| 5. Reject | v=DMARC1; p=reject; rua=... | Refuse the message | Keep monitoring |
- Start at
p=none. RFC 9989 calls this Monitoring Mode and saysp=nonemust not change existing mail handling. There’s no downside to turning it on immediately once you have anruaaddress you’ll actually monitor. - Review reports for every legitimate source. Google suggests reviewing daily and says one week is usually representative. RFC 9989 warns it can take many months for domains that send infrequently, so watch for annual or seasonal senders. For an online shop, email authentication for online stores lists the senders a store usually has, so you know which sources to look for.
- Fix each failing legitimate source. RFC 9989 says these must be fixed before you publish an enforcing policy. Usually that means a missing SPF
include:or a service that isn’t DKIM-signing with your domain. Re-check with DMARC check and MX lookup after each fix. - Step up gradually. Under RFC 7489,
pctapplied a policy to a share of failing mail, and Google’s guide still showsp=quarantine; pct=5. RFC 9989 listspctas historic, because values other than 0 and 100 were applied inaccurately, and replaces it witht=y. Both RFC 7489 and RFC 9989 tell receivers to ignore tags they don’t recognise, so a receiver that only implements RFC 7489 ignorest=yand applies the full policy. Watch the reports at each stage either way. - Switch to
p=rejectonce reports consistently show zero unexplained failures from legitimate sources across at least one full reporting cycle. AdminKit’s DMARC generator builds the record for each stage.
Skipping this review is how a rollout loses legitimate mail. RFC 7489 says rejection should happen during the SMTP transaction, so the receiver never accepts the message, and any bounce goes to the sending system. For a third-party platform, that is often not you.
How do subdomains and test mode show up in reports?
Subdomains are reported under the policy that covers them, so check the header_from value in each record, not only the IP. RFC 9990 §3.1.2 says a receiver generates a separate report for each DMARC policy domain it encounters. Its example: if example.com and bar.example.com both publish records but foo.example.com does not, mail from foo.example.com appears as its own record inside the example.com report.
That matters for rollout because a subdomain without its own record follows your organizational domain’s sp value, or p if sp is absent. RFC 9989 adds np for mail from subdomains that don’t exist in DNS at all. Before moving the apex to reject, look for subdomain rows from legitimate senders, such as a ticketing system sending as support.example.com.
Test mode leaves a trace too. When t=y stops a receiver from applying your policy, RFC 9990 defines the override reason policy_test_mode for the record’s reason element. Seeing it confirms the receiver read your test flag. If failing mail from the same receiver shows quarantine or reject with no such reason, that receiver applied the full policy.
Finally, size. RFC 7489 says reports should be gzip-compressed, noting that a commonly observed receiver limit is ten megabytes, and that a receiver which can’t deliver a report to any listed address should send a short error report instead. If reports from a large receiver stop arriving, check that your report mailbox accepts large attachments.
Why keep monitoring after p=reject?
Because your senders change and a reject policy hides the fallout. A new marketing tool or a migrated mail platform that isn’t authorized will simply fail under p=reject, and the aggregate reports are often the first place it shows up. Treat the rua address as a standing operational responsibility, not a one-time rollout step.
Two details also show up in reports at this stage. RFC 9989 lets receivers accept failing mail despite p=reject, for example from mailing lists, and asks them to say so with a policy override reason. And RFC 7489 notes that a report spanning a policy change may mix old and new policies. AdminKit for email deliverability teams covers the fuller recurring checklist, SPF, DKIM, DMARC, MX and reverse DNS, worth running alongside that ongoing report review. This rollout process is part of AdminKit’s email authentication cluster, which covers SPF and DKIM setup alongside the DMARC reporting discussed here.
This guide is general information, not a security audit of your specific network — see our disclaimer for what AdminKit's guidance does and does not commit to.
Frequently asked questions
What's the difference between rua and ruf reports?
`rua` (aggregate reports) are XML summaries, normally daily, of pass/fail counts grouped by sending source — the main tool for rollout decisions. `ruf` (failure, or forensic, reports) describe individual failing messages and, as RFC 9991 notes, can contain header fields or even the whole message, which may include personal data. Many receivers don't send them; Google's own DMARC documentation states that Gmail doesn't support the ruf tag. Don't rely on ruf for rollout decisions.
How long should I stay at p=none before moving to quarantine?
Long enough to see every legitimate sending source at least once, including ones that don't send daily — payroll providers, annual-report tools, or seasonal marketing platforms. Google's recommended rollout treats one week of daily reports as usually enough to be representative of your mail streams; RFC 9989 warns that, depending on how often you send, it may take many months before you're sure all of your mail authenticates. A domain with infrequent senders should wait for those senders to appear in the reports.
A report shows a failing source I don't recognize — what do I do?
Identify it first: the report's `source_ip` and `header_from` fields, cross-referenced with your own list of approved sending services, usually reveal whether it's a forgotten legitimate sender or genuine spoofing. If you can't identify it and it's failing both SPF and DKIM, it's very likely not you — which is exactly what DMARC enforcement is meant to stop.
Can I go straight from p=none to p=reject?
You can, but it's risky — no DMARC specification requires a staged rollout, but skipping quarantine means any sender you failed to identify in the aggregate reports gets outright rejected instead of just flagged, with no safety net. RFC 7489's `pct` tag let you apply a new policy to a percentage of failing mail first. RFC 9989, which replaced RFC 7489 in May 2026, removed `pct` because values other than 0 and 100 were applied inaccurately, and added `t=y` (test mode), which asks receivers to apply one level below the published policy while you watch the reports.
Do aggregate reports include message content?
No. Aggregate reports contain only metadata — source IP, volume, pass/fail counts, and policy applied — not subject lines or message bodies, which is part of why most receivers are willing to send them while fewer send the more detailed failure (ruf) reports.
Can reports go to a mailbox at a different domain, such as a DMARC service?
Yes, but the receiving domain has to agree. RFC 7489 §7.1 and RFC 9990 §4 require a receiver to look for a TXT record at <your domain>._report._dmarc.<report domain> containing v=DMARC1 before sending reports to an address outside your organizational domain. Without that record, the external address is ignored and no reports arrive there.
Sources
- RFC 7489 — Domain-based Message Authentication, Reporting and Conformance (DMARC)
- dmarc.org — DMARC overview
- IETF Datatracker — RFC 7489, DMARC (canonical document record)
- RFC 9989 — Domain-Based Message Authentication, Reporting, and Conformance (DMARC)
- RFC 9990 — DMARC Aggregate Reporting
- RFC 9991 — DMARC Failure Reporting
- Google Workspace Admin Help — Recommended DMARC rollout
- Google Workspace Admin Help — About DMARC reports
- Google Workspace Admin Help — Set up DMARC
Related tools
Want AdminKit to watch this for you?
Free checks tell you the state today. Protective DNS and monitoring tell you when it changes.
Talk to an expert
Huy Nguyen (David), Founder & CEO, Netbase JSC
Ask whether this fix applies to your setup, or get a second pair of eyes on your DNS and mail records.