Blazalek.com

When to move DMARC from p=none to quarantine and reject

A clean DMARC dashboard doesn’t prove your domain is ready for quarantine. Before moving past monitoring, you need every approved sending path identified, known authentication gaps repaired, and an actionable fallback ready in case legitimate messages get blocked. Here is how to weigh the evidence you have, and how to back out and recover mail if enforcement harms legitimate messages.

By

On this page

Move DMARC from p=none to quarantine only when three things are true: you have identified every legitimate sending path, you have fixed every known authentication misalignment, and you have staged a monitored rollout with an immediate fallback record. A clean reporting dashboard does not prove production readiness on its own. The recommended operational progression moves to quarantine first and evaluates reject as a distinct decision later. That sequence reflects practical risk management rather than a rigid RFC timeline.

Define what the policy change covers

Start with your sender inventory, capturing direct user email, automated alerts, forwarding, mailing lists, and infrequent batch jobs. Identify the DNS operator and the person verifying business mail. Save the current TXT record, its TTL, the planned change time, and a less restrictive fallback record.

Map each stream to the visible From domain, its DMARC policy, the apex organizational policy, and active overrides. A change at the apex can reach subdomains through inheritance, and under RFC 9989, subdomains inherit your apex policy p by default (unless np applies), so omitting sp does not automatically leave subdomains unprotected. Verify these rules against the defined policy tags before modifying DNS.

Working SPF, valid DKIM, and aggregate report ingestion must precede monitoring with p=none and rua; RFC 9989 classifies them as deployment prerequisites. If that initial record is still ahead of you, consult how to roll out DMARC without breaking your email. Collecting reports you never read won’t help you. When sending reports externally, confirm the destination published the required destination authorization. Receivers have no obligation to send reports (reporting participation), so missing reports call for a look at collection and coverage; they do not tell you whether traffic is healthy or absent.

Do clean DMARC reports prove you are ready?

For each approved path, keep three layers distinct: raw SPF and DKIM results, identifier alignment, and receiver disposition. DMARC passes only when a passing SPF or DKIM identifier aligns with the visible From domain under the DMARC pass rule; a correct configuration proves nothing about a particular message. When alignment fails, follow the alignment diagnosis procedure.

Match observed reports against expected sending cadences. Assessing low-frequency streams takes time: a quarterly digest stays unassessed until it runs, regardless of how cleanly daily mail passes. That is why enforcement decision guidance emphasizes cadence. Every legitimate authentication defect must be remediated before enforcement.

When analyzing reports, examine the reporting organization, date range, IP path, and evaluated policy. A report that spans a change may record only the final configuration, which limits before-and-after attribution. Receiver disposition codes also do not measure inbox placement. Corroborate unexpected failures with server logs and raw samples, as connecting IPs do not identify application accounts. RFC 9990 details what aggregate evidence contains and its placement limits.

Inspect raw headers from authorized mailboxes, trusting Authentication-Results only when you trust both the reporting server and the delivery path. An attacker can forge an authserv-id header on untrusted hops, and one sample covers only that specific path (authentication-result trust, RFC 8601 section 7.1). Failure reports are a separate, optional mechanism; they may expose message content and are not required here (failure reporting).

Synthetic example: equipment bookings

This fictional scenario tracks From: bookings@equipment.example used for transactional receipts and a quarterly digest reporting overdue equipment. It illustrates diagnostic reasoning rather than live production traffic.

What you observeWhat it meansWhat to do
Receipts validate with DKIM d=equipment.example; the digest has not run yet.Receipts pass; the digest’s readiness is unknown.Exercise the approved digest path or wait for its representative event; check authentication and the destination outcome.
The digest’s SPF passes for the MAIL FROM domain vendor.example.net, and its only DKIM signature validates with d=vendor.example.net.Neither identity aligns with equipment.example, so DMARC fails.Repair the digest configuration before enforcement.
Later, representative results satisfy the readiness checks.Quarantine becomes a candidate, with confidence limited by what you have actually observed.Prepare the change and the fallback.
An approved path starts failing DMARC after the change.Progression must stop; the cause and the delivery impact need investigation.Check the applicable policy and the outcomes; restore the fallback if enforcement is harming legitimate mail.

Moving from none to quarantine, and deciding on reject

Advance through the operational gates below once all discovered misalignments are fixed. Quarantine is already enforcement. Neither quarantine nor reject guarantees a specific folder or gateway action: mailbox providers retain local handling discretion, and clean authentication is no inbox guarantee.

StageEvidence that lets you proceedStop condition
none → quarantineExpected paths assessed; known legitimate gaps repaired; reporting and fallback ready.An expected path is still unobserved, or a legitimate failure is unresolved.
Observe quarantineApproved paths still authenticate, and delivery evidence shows no new DMARC-related harm.An unexplained failure in an approved stream, confirmed harm, or expected reporting that has gone missing.
Consider rejectQuarantine evidence remains satisfactory; the DKIM and indirect-flow conditions below are addressed.Reliance on SPF alone, or an unresolved interoperability risk.

When considering p=reject, RFC 9989 explicitly requires valid DKIM signatures and forbids relying on SPF alone. That is a requirement for deploying reject; it does not change the pass rule for an individual message, where either SPF or DKIM can still produce the pass. Furthermore, DKIM cannot guarantee that a signature survives message modification by intermediaries such as mailing lists.

The specification advises against setting reject on domains whose users participate in discussion lists. If your domain still targets reject, observe traffic for at least a month at none followed by an equivalent observation period at quarantine. Compare the dispositions, and restrict list posting and/or inform users. Those are scoped recommendations under the interoperability conditions, not a universal timetable and not proof of readiness; you still need to fix existing authentication failures before turning on quarantine.

Do not use percentage-based rollout recipes: RFC 9989 removed pct. Under the test-mode definition, setting t=y converts reject to quarantine and quarantine to none, while having zero effect on none, meaning it cannot serve as an all-purpose safety throttle. The May 2026 standard does not prove receiver adoption: Google’s setup page, updated August 26 and checked September 9, 2026, still references pct. That mismatch in documentation does not tell you how the implementation behaves.

Rolling back when enforcement breaks legitimate traffic

If metrics or user reports show legitimate business email is failing, stop your rollout immediately. Correlate the published DNS record, recent authentication verdicts, and bounce reports. If enforcement is damaging legitimate email, roll back to your prepared fallback record, and maintain aggregate reporting throughout the rollback. Stepping back to quarantine offers no relief if quarantine itself is causing the delivery failures.

Confirm the rollback across authoritative DNS nameservers, test mailboxes, and incoming aggregate reports. A saved control-panel change or a single DNS answer cannot establish recovery everywhere. TTL defines cache lifetime, but it doesn’t guarantee instant global propagation, and report periods delay visibility further.

Messages rejected during enforcement need a separate review. When a receiving server permanently rejects a transaction at the end of SMTP DATA, it must not deliver that message later; responsibility stays with the client, which should not retry that server without review and intervention (SMTP rejection semantics, RFC 5321 section 4.2.5). Updating DNS records does not resend lost traffic. Check mail queues before retrying failed messages, and only resume your rollout once you’ve found and fixed the root cause.

Frequently asked questions

Do clean DMARC aggregate reports prove a domain is ready for quarantine?

No. Clean reports only reflect active traffic during the reporting window, meaning infrequent senders like quarterly digests stay unassessed until they run. You must account for all approved sending paths before enabling quarantine.

Does publishing p=quarantine or p=reject determine exact receiver placement?

No. Mailbox providers retain local handling discretion under RFC 9989, and passing authentication does not guarantee inbox delivery. Enforcement tells receivers your policy preference, not their final folder placement.

Can you advance to p=reject using SPF alone without DKIM?

No. RFC 9989 explicitly requires valid DKIM signatures and forbids relying solely on SPF when deploying reject. While an individual message can pass DMARC via SPF, advancing the domain policy to reject requires DKIM.

Should you use the pct tag to roll out quarantine gradually?

No. RFC 9989 removed the pct tag entirely from the DMARC specification. Its test mode lowers reject to quarantine and quarantine to none, and t=y has no effect on none, so don’t treat it as a universal safety switch.

How do you recover if DMARC enforcement begins blocking legitimate email?

Restore the less restrictive fallback DNS record immediately while keeping aggregate reporting active. Rolling back DNS will not re-deliver messages rejected at SMTP DATA. Inspect message state before you arrange recovery.

Email deliverability consultant

Get a second opinion on your email setup

Tell me what your customers are seeing or what you’re planning to change. In 30 minutes, we’ll work out what your team can check first and whether a deliverability audit would help.

I led deliverability at GetResponse, designed a dedicated newsletter system for Easy.Tools and helped Youmoney migrate its automated emails in stages. Read the case studies

Book free call

Email deliverability audit

See what I check and read a sample report with recommendations for your team.