Domain health guide
DMARC policy: none, quarantine, or reject?
DMARC policy tells receiving mail systems what to do when a message fails DMARC. Start by understanding your legitimate senders before moving from monitoring to enforcement.
What each DMARC policy does
A DMARC record publishes a policy at the domain level. With p=none, a receiving system is asked to monitor failures without applying enforcement. With p=quarantine, failing mail is treated as suspicious. With p=reject, the receiving system is asked to reject failing mail.
The policy is not a guarantee that every receiver will behave identically. It is a public instruction that works alongside SPF and DKIM alignment checks.
Why p=none is not the same as protection
p=none is useful while you learn which systems legitimately send mail for the domain. It can help prevent an enforcement policy from disrupting valid messages too early.
Because it does not request quarantine or rejection, p=none should be treated as a monitoring stage rather than the final protection level. CloudSpex reports p=none as a warning so it is not mistaken for full enforcement.
A safe progression
Inventory every legitimate sender first: your mailbox provider, support desk, marketing platform, transactional email service, and any application that sends on your behalf. Confirm that each source has the intended SPF and DKIM configuration.
Review DMARC reports or your mail-provider evidence, then move in small steps. A common sequence is p=none, then p=quarantine, then p=reject once legitimate traffic is understood. Do not copy a policy from another domain without checking its sending services.
- Check whether a DMARC record is present and which policy it publishes.
- Confirm SPF and DKIM are configured for legitimate senders.
- Review legitimate mail before changing enforcement.
- Publish a stricter policy only after the sending inventory is understood.
- Recheck the public record after DNS changes propagate.
Common mistakes
Treating a syntactically present record as complete protection is a common mistake. A p=none policy is intentionally weaker than quarantine or reject.
Another mistake is enabling reject while an overlooked service still sends unauthenticated mail. That can interrupt wanted mail, so ownership and change control matter as much as the DNS record itself.
What the three records look like
A DMARC record is a single TXT record published at _dmarc followed by the domain name. The first two tags must be v and p, in that order. Everything else is optional, but a record without a rua address gives you a policy and no way to see its effect.
The three records below are the same domain at three stages. Only the p tag changes between the first and the last; the pct tag in the middle one is the tool for moving gradually.
; Stage 1 - monitor only
_dmarc.example.com. TXT "v=DMARC1; p=none; rua=mailto:dmarc@example.com"
; Stage 2 - quarantine a quarter of failing mail
_dmarc.example.com. TXT "v=DMARC1; p=quarantine; pct=25; rua=mailto:dmarc@example.com"
; Stage 3 - reject all failing mail
_dmarc.example.com. TXT "v=DMARC1; p=reject; rua=mailto:dmarc@example.com"How a receiver decides whether a message fails
The policy only applies to messages that fail DMARC, so it is worth being exact about what failing means. A message passes DMARC when at least one of two things is true: SPF passed and the domain SPF checked is aligned with the domain in the visible From address, or DKIM passed and the signing domain is aligned with that same From domain.
Aligned means the same organisational domain in relaxed mode, which is the default, or exactly the same domain in strict mode. A message from news.example.com signed by example.com is aligned in relaxed mode and not aligned in strict mode.
This is why a service can pass SPF and DKIM and still fail DMARC. A newsletter platform that signs with its own domain and uses its own bounce address produces a valid signature and a valid SPF result, and neither of them belongs to your domain. The fix is always at the sending service: turn on a custom signing domain, a custom return path, or both.
What changes for the recipient at each stage
With p=none, a failing message is handled as if no DMARC record existed. The receiver may still filter it for its own reasons, but your policy does not ask for anything. With p=quarantine, the receiver is asked to treat a failing message as suspicious, which in practice usually means the spam folder. With p=reject, the receiver is asked to refuse the message during delivery, so the sender gets a bounce and the recipient sees nothing.
Quarantine is recoverable and reject is not. A legitimate message that lands in spam can be found; a rejected message never arrives. That difference is the reason to pass through quarantine instead of jumping from none to reject.
Receivers treat the policy as a request. Most large mailbox providers honour it, but each applies local rules as well, and a small share of failing mail may be handled differently from what the policy asks.
A staged rollout you can follow
The schedule below is a pattern, not a rule. The condition for each step is the same: the aggregate reports show that the mail still failing comes only from sources you have decided not to authorise.
- Publish p=none with a rua address and collect reports for at least two weeks, long enough to include monthly senders such as invoicing runs.
- List every source in the reports. For each legitimate one that fails, fix SPF or DKIM alignment at that service.
- Move to p=quarantine with pct=25. Watch the reports and your support inbox for a week.
- Raise pct to 50, then 100, leaving at least a few days between changes.
- Move to p=reject with pct=100 once quarantine at 100 percent has produced no legitimate failures.
- Keep the rua address after reaching reject. New senders are added to a domain all the time, and the reports are how you find out.
Subdomains and the sp tag
A DMARC record on the organisational domain also covers its subdomains unless a subdomain publishes its own record. The sp tag lets you set a different policy for subdomains than for the domain itself.
Two uses are common. A domain that is still at p=none can publish sp=reject if it sends no mail from subdomains at all, which closes off an easy route for spoofing. In the other direction, a domain at p=reject can publish sp=none while a subdomain used by a marketing platform is still being fixed. Without the sp tag, subdomains inherit p.
_dmarc.example.com. TXT "v=DMARC1; p=none; sp=reject; rua=mailto:dmarc@example.com"Domains that send no mail
Parked domains, redirect domains and old brand names are favourite targets for spoofing precisely because nobody watches them. They need no rollout: there is no legitimate mail to break. Publish p=reject immediately, together with an SPF record that authorises nothing.
example.org. TXT "v=spf1 -all"
_dmarc.example.org. TXT "v=DMARC1; p=reject; rua=mailto:dmarc@example.com"When forwarding and mailing lists break DMARC
Two ordinary situations cause legitimate mail to fail. When a recipient forwards mail automatically to another mailbox, the forwarding server becomes the sender as far as SPF is concerned, so SPF fails. DKIM usually survives forwarding because the signature travels with the message, which is the main reason to make sure every sender signs with an aligned DKIM key rather than relying on SPF alone.
Mailing lists are harder. A list that adds a footer or changes the subject line alters the signed content, so the DKIM signature no longer verifies. Many lists now rewrite the From address to avoid this. In aggregate reports these cases appear as failures from servers you do not recognise, and they are a reason to read the reports carefully before concluding that an unknown source is abuse.
Checking the result
After each change, read the published record back rather than trusting the DNS panel. A second DMARC record left over from an earlier attempt makes receivers ignore both. A missing semicolon or a tag in the wrong order can make the whole record invalid, which silently returns the domain to no policy at all. The CloudSpex DMARC Checker reads the record a receiver would see, reports the policy for the domain and its subdomains, and says when a part of the record cannot be interpreted.
Telling people before you tighten
A policy change is a DNS edit that affects other teams' mail. Before moving to quarantine or reject, tell whoever runs marketing, support and finance systems which date the change takes effect and what a failure would look like: a campaign with unusually low opens, or customers saying a message went to spam. Most problems after a policy change are reported by a colleague, not by a tool, and a colleague who was warned reports them on the first day.
Frequently asked questions
Is p=none a valid DMARC policy?
Yes. It is valid for monitoring, but it does not ask receivers to quarantine or reject failing mail.
Which DMARC policy is strongest?
p=reject is the strongest of the three policy values, but it should be used only after legitimate senders have been checked.
Does CloudSpex change my DMARC record?
No. CloudSpex reads public DNS signals and does not change DNS or email settings.
How long should a domain stay at p=none?
Long enough for the aggregate reports to cover every regular sender, including ones that run monthly. For most domains that is two to four weeks. Staying at p=none indefinitely leaves the domain unprotected.
Does pct work with p=none?
No. pct only has an effect with quarantine or reject. With p=quarantine, messages outside the percentage are treated as p=none; with p=reject, they are treated as quarantine.
Can I skip quarantine and go straight to reject?
You can, and for a domain that sends no mail you should. For a domain with real mail, quarantine is the stage where a mistake costs a message in the spam folder instead of a message that never arrives.
Use public checks as a starting point
CloudSpex reads limited public-facing signals. It does not change DNS, website, certificate, or email settings. Confirm the responsible provider and test relevant changes before applying a remediation.
Editorial details
Author and reviewer: CloudSpex. Sources: public technical standards and provider documentation relevant to the topic. Last reviewed: November 10, 2026. We update guides when the product behavior or the underlying guidance changes.