Moving DMARC from p=none to p=reject without losing mail

A step-by-step path from monitoring to full protection, and how to know when each step is safe.

p=none only watches. Forged mail using your domain is still delivered. The protection starts at p=quarantine (failing mail goes to spam) and is complete at p=reject (failing mail is refused). The risk is blocking your own mail from a service you forgot about. Here's how to get there without that happening.

1. Collect reports for a few weeks

Start with p=none and an rua= address. Give it at least two to four weeks, long enough to see monthly mail like invoices and newsletters. See reading DMARC reports.

2. Fix every legitimate sender

Go through every source in the reports. For each one you recognise, make sure DKIM passes with your domain (aligned). For sources you don't recognise, find out: ask colleagues, check the IP owner. Leave forwarding and mailing-list traffic aside; it will always fail partly.

You're ready for the next step when practically all mail from your own services passes, consistently, for a couple of weeks.

3. Quarantine

v=DMARC1; p=quarantine; rua=mailto:…

Watch the reports and ask colleagues to tell you if anything lands in spam. If you want to go more gradually, the older pct= tag (for example pct=25) applies the policy to only part of the failing mail; not every receiver honours it, and the updated DMARC specification replaces it with a simple testing flag (t=y).

4. Reject

v=DMARC1; p=reject; rua=mailto:…

After a few quiet weeks at quarantine, move to reject. Keep reading the reports: new services get connected all the time, and with reject their mail will bounce until they're set up.

Don't forget subdomains

sp=none leaves every subdomain open to forgery while the main domain is protected. Leave sp out (it then follows p) unless a subdomain really needs a different policy. Also protect the domains you don't send from: they can go to p=reject straight away.