schedule 18-min read

How to Check Your DMARC Policy

DMARC AI editorial team · Last updated

How to check your DMARC policy: query _dmarc with dig, find which record covers a subdomain, then confirm the applied policy in headers and reports.

01

How do you check your DMARC policy?

To check your DMARC policy, run three tests: look up the TXT record at _dmarc.yourdomain and read its p, sp, np and t tags; work out which record actually covers the domain or subdomain you send from; then send a real message and confirm the policy receivers applied, in the Authentication-Results header and in aggregate reports.

A policy can look right in DNS and still not be in force, for example because a second record cancels it or t=y softens it. This guide covers the commands, the RFC 9989 rules, what to read in Gmail and Outlook, and a routine for MSPs (managed service providers) checking a whole portfolio. For the concepts behind DMARC (Domain-based Message Authentication, Reporting and Conformance), start with the DMARC guide.

02

Key takeaways

  • A DMARC policy is in force only when there is exactly one valid record, the right record covers the sending domain, and receivers report the expected disposition.
  • Read p, sp, np and t together: p=reject; t=y asks receivers to quarantine, not reject.
  • Under RFC 9989, a subdomain without its own record is covered by its Organizational Domain's record, found by a DNS tree walk, not by the Public Suffix List.
  • Confirm the policy with real evidence: the dmarc= result in a message header, and the policy_published and disposition values in aggregate reports.
  • pct, ri and rf are historic in RFC 9989. Do not rely on a leftover percentage; remove it.
03

What are the three checks for a DMARC policy?

The three checks are a DNS check, a coverage check and an evidence check. Run them in order; the table shows what each proves and what it cannot.

Check What you test Tool Proves Does not prove
1. Published record TXT at _dmarc.domain dig, nslookup, Resolve-DnsName What policy you published, and whether the record is valid That it covers the subdomain you send from
2. Coverage Which record applies to a subdomain dig per label, RFC 9989 Section 4.10 Which record and which tag (p, sp or np) a receiver will use What receivers actually did
3. Evidence Authentication-Results header, aggregate reports Gmail Show original, Outlook message details, rua reports The policy receivers applied to real mail Nothing more is needed for the policy

A free DMARC checker covers whether a record exists; this page is about the policy actually in force.

04

Check 1: what DMARC policy is published in DNS?

The published policy is the p= value in the single TXT record at _dmarc.yourdomain that starts with v=DMARC1. Query that exact name, confirm one record comes back, and read the policy tags together, because sp, np and t change what p means in practice.

How do you look up a DMARC record with dig, nslookup or Resolve-DnsName?

Query the TXT record at _dmarc plus the domain. On macOS or Linux use dig; on Windows use nslookup or PowerShell's Resolve-DnsName. The PowerShell and nslookup forms below are the ones Microsoft gives in its DMARC configuration guide.

# macOS / Linux
dig +short TXT _dmarc.example.co.uk

# Ask a public resolver to avoid a stale local cache
dig +short TXT _dmarc.example.co.uk @1.1.1.1

# Windows
nslookup -type=TXT _dmarc.example.co.uk
Resolve-DnsName -Name "_dmarc.example.co.uk" -Type TXT | Select-Object -ExpandProperty Strings

A healthy answer is one line:

"v=DMARC1; p=reject; sp=reject; np=reject; rua=mailto:dmarc@reports.example.net"

If the authoritative server (find it with dig +short NS example.co.uk) shows the new value but a public resolver does not, you are waiting for a TTL (time to live) to expire. For record syntax tag by tag, see what is a DMARC record and how do you read it.

Which tags decide the DMARC policy in force?

Four tags decide the policy: p for the domain itself, sp for existing subdomains, np for non-existent subdomains, and t for test mode. Defaults apply when a tag is absent.

Tag What it controls (RFC 9989 Section 4.7) Default if absent What to check
p Policy for the domain: none, quarantine or reject Treated as none in an otherwise valid record Is it the level you think you are at?
sp Policy for existing subdomains Value of p Is it weaker than p by accident?
np Policy for non-existent subdomains sp, or p if there is no sp Set np=reject to stop made-up subdomains
t Test mode: t=y applies one level softer n Is t=y still there after a rehearsal?
adkim / aspf Strict (s) or relaxed (r) alignment r Strict fails senders that sign or bounce from a subdomain
rua Where aggregate reports go No reports Is there at least one valid mailto: address?

RFC 9989 Section 9.3 lists pct, rf and ri as "historic", and Appendix A.6 explains that t replaces the useful part of pct. RFC 9989 does not define how receivers treat a leftover pct value, and some still act on it (see the Microsoft action=pct.* values below), so remove it rather than rely on it. The new DMARC tags np, psd and t covers each one in depth.

What does t=y do to the policy?

t=y tells receivers to apply the policy one level below what p says. RFC 9989 defines it so that p=quarantine with t=y is handled as none, and p=reject with t=y is handled as quarantine. It is a rehearsal switch, not a permanent setting.

So these two records enforce differently:

"v=DMARC1; p=reject; t=y; rua=mailto:dmarc@reports.example.net"   -> failing mail is quarantined
"v=DMARC1; p=reject; rua=mailto:dmarc@reports.example.net"        -> failing mail is rejected

A forgotten t=y explains a domain "at reject" that rejects nothing.

What happens if the record has an error?

RFC 9989 Section 4.10 says that if more than one DMARC record is returned for a name, "they are all discarded", so two records mean no policy at all. Section 4.10.1 says that if p is missing or invalid but rua holds a valid address, receivers act as if p=none was published; without a valid rua, they apply no DMARC processing.

Other syntax errors fall back to defaults (Section 4.8), so a typo such as sp=rejct silently becomes the value of p. Why your DMARC record is not working walks through the failure modes in troubleshooting order.

05

Check 2: which DMARC record applies to a subdomain?

A subdomain with its own _dmarc record uses that record's p value. A subdomain without one is covered by its Organizational Domain's record, which receivers find with the DNS tree walk in RFC 9989 Section 4.10, and the policy then comes from sp if the subdomain exists or np if it does not.

How does the RFC 9989 tree walk find the record?

The receiver first queries _dmarc at the exact From domain. If no valid record is there, it removes labels from the left and queries each parent in turn, up to a limit of eight queries. The Organizational Domain is the highest name in the tree with a DMARC record, unless a record with a psd tag stops the walk earlier.

RFC 9989 Appendix C.3 says the tree walk replaces the Public Suffix List approach of RFC 7489. Two consequences matter:

  • A record on an intermediate name does not cover names below it by default. In the RFC's own example (Appendix B.4.2), signing.example.com has a record, but the walk continues to example.com, the highest name with a record. So _dmarc.shop.example.co.uk does not set the policy for mail.shop.example.co.uk; the sp value at example.co.uk does. A record containing psd=n stops the walk at that name.
  • The exact match always wins. If a subdomain needs its own policy, publish _dmarc at that exact name.

Check it with one lookup per label:

dig +short TXT _dmarc.mail.shop.example.co.uk   # exact From domain first
dig +short TXT _dmarc.shop.example.co.uk
dig +short TXT _dmarc.example.co.uk
dig +short TXT _dmarc.co.uk                     # normally empty

From domain Records found Record used Tag used
example.co.uk _dmarc.example.co.uk example.co.uk p
news.example.co.uk (has its own record) _dmarc.news.example.co.uk news.example.co.uk p of that record
mail.example.co.uk (exists, no record) Only _dmarc.example.co.uk example.co.uk sp
invoices.example.co.uk (does not exist) Only _dmarc.example.co.uk example.co.uk np, else sp, else p
mail.shop.example.co.uk (no record) _dmarc.shop.example.co.uk and _dmarc.example.co.uk example.co.uk (highest record) sp

To find out whether a subdomain "exists" for the np rule, query it: dig mail.example.co.uk returning NXDOMAIN means the name does not exist. DMARC for subdomains and non-existent subdomains and the np tag cover the policy design side.

When does the psd tag matter?

For most organisations, never. psd=y is for operators of public suffix domains, such as a registry (RFC 9989 Section 5.2), and psd=n declares that a record belongs to an Organizational Domain (Section 11.8). A stray psd=y on an ordinary company domain is a finding to fix. DMARCbis and public suffix domains explains the background.

06

Check 3: what policy do receivers actually apply?

Receivers record their verdict in the Authentication-Results header of each message and summarise it in aggregate reports. Send a real message to an external mailbox, read the dmarc= result, then confirm across all mail in the reports.

RFC 9989 Section 9.2 registers five results for the dmarc method: pass, fail, none (no DMARC record exists), temperror and permerror. It also registers a policy.dmarc property for the evaluated policy, which some receivers add after the result.

How do you check DMARC in Gmail?

Send a message from the domain to a Gmail account and open its full header. Google's header instructions say: in a browser, next to Reply, click More, then Show original. The full header opens in a new window, with Copy to clipboard.

Find the Authentication-Results line added by Google and read the dmarc= value and header.from= domain; only the receiver's verdict counts. The email header analyzer lays the fields out, and how to read email headers explains the rest of the header.

How do you check DMARC in Outlook and Microsoft 365?

In new Outlook, open the message, select More actions, then View > View message details, as Microsoft Support describes. Exchange Online writes the DMARC result together with an action= value that names the policy it found and acted on.

Authentication-Results: spf=pass (sender IP is 198.51.100.10)
  smtp.mailfrom=bounces.adatum.com; dkim=pass (signature was verified)
  header.d=adatum.com; dmarc=fail action=oreject
  header.from=contoso.com;compauth=fail reason=000

In this example from Microsoft's DMARC guide, SPF and DKIM pass for adatum.com but the From domain is contoso.com, so DMARC fails on alignment and p=reject applies. Microsoft documents these values:

Header value What Microsoft says it means
dmarc=pass action=none DMARC passed
dmarc=fail action=none Sender published p=none; no action taken
dmarc=fail action=quarantine Sender published p=quarantine; message quarantined or junked
dmarc=fail action=oreject Sender published p=reject; message rejected
action=pct.quarantine / pct.reject A pct value below 100 was published (a tag RFC 9989 removed)
dmarc=bestguesspass No DMARC record exists; Microsoft infers it would pass

Handling also depends on the receiving tenant's Honor DMARC record policy setting, so a delivered message with action=oreject points to that tenant, not your record. DMARC reports say reject but email was delivered covers that case.

How do aggregate reports confirm the policy in force?

Every aggregate report repeats the record the receiver used in its policy_published block and states, per sending source, the disposition it applied. Compare policy_published with what you meant to publish, then check disposition for failing sources across all reported mail.

RFC 9990 Section 3.1.1.5 defines policy_published with the policy domain, discovery_method (psl or treewalk), p, sp, np, adkim, aspf, fo and testing (the value of t).

<policy_published>
  <domain>example.co.uk</domain>
  <discovery_method>treewalk</discovery_method>
  <p>reject</p>
  <sp>reject</sp>
  <np>reject</np>
  <testing>y</testing>
</policy_published>
...
<record>
  <row>
    <source_ip>203.0.113.25</source_ip>
    <count>38</count>
    <policy_evaluated>
      <disposition>quarantine</disposition>
      <dkim>fail</dkim>
      <spf>fail</spf>
      <reason>
        <type>policy_test_mode</type>
      </reason>
    </policy_evaluated>
  </row>
  <identifiers>
    <header_from>example.co.uk</header_from>
  </identifiers>
</record>

Here the receiver found p=reject with testing y, so it quarantined 38 failing messages, citing policy_test_mode. RFC 9990 defines the disposition values as none (no action taken), pass (passing DMARC with an enforcing policy), quarantine and reject, and the override reasons as local_policy, mailing_list, trusted_forwarder, policy_test_mode and other. If policy_published shows a different domain from the one you expected, the tree walk landed somewhere you did not intend. DMARC aggregate reports explained walks through the rest of the format.

07

What does each DMARC result mean?

Each result points to a specific next step. Match the header or report value to a row, then fix the record or the sender it names.

What you see What it means What to do
dmarc=none in a header The receiver found no DMARC record for the From domain Publish a record at _dmarc; see how to fix "no DMARC record found"
dmarc=permerror The record could not be used, for example badly formed Check syntax and make sure only one _dmarc TXT record exists
dmarc=fail action=none Mail failed, but p=none asked for no action Fix the failing sender, then move policy up when reports allow
dmarc=fail, disposition one level softer than p t=y is in the record Remove t=y once the rehearsal is over
policy_published domain is the parent, not the subdomain The subdomain has no record and inherits sp or np Confirm that is intended, or publish a subdomain record
disposition none with reason mailing_list or local_policy The receiver chose to override your policy Often forwarding; check whether DKIM survives the path
08

What are the common DMARC policy problems?

These six problems make a policy look right without being in force. Each is visible with a dig lookup or one aggregate report, so check all six when you take over a domain.

  1. No record. dig +short TXT _dmarc.example.co.uk returns nothing, and headers show dmarc=none. Publish one, starting at p=none with rua if the domain sends mail; how to create a DMARC record has the steps.
  2. Multiple records. Two lines start with v=DMARC1, often an old one plus a vendor's. RFC 9989 discards both, so the domain has no policy. Merge them into one.
  3. Record on the wrong name. The string sits at the apex, at dmarc.example.co.uk without the underscore, or at _dmarc.example.co.uk.example.co.uk because the DNS panel appended the zone twice.
  4. Syntax error. A missing v=DMARC1 at the start, a misspelt value or smart quotes from a word processor.
  5. p=none forgotten. Published for monitoring and never moved, so reports show disposition none indefinitely. Fix senders, then move to p=quarantine and p=reject, following how to move from monitoring to enforcement safely.
  6. t=y left on. The record says p=reject but every receiver applies quarantine, and reports show testing y. Remove the tag.

A seventh: rua on another organisation's domain without its authorisation record, so no reports arrive for Check 3. Authorising external DMARC report destinations explains the fix.

09

How should an MSP check DMARC policy across a client portfolio?

At portfolio scale, script the DNS check across every client domain and let aggregate reports do the evidence check, because a test message per domain does not scale. Run the routine monthly or quarterly, and after every DNS change window.

A short loop gives a policy inventory from a plain list of domains:

while read d; do
  r=$(dig +short TXT "_dmarc.$d" | tr -d '"' | grep -i '^v=DMARC1')
  n=$(printf '%sn' "$r" | grep -c .)
  echo "$d | records=$n | $r"
done < client-domains.txt

Then work through the output:

  1. Flag records=0 and records=2 or more. Both mean no policy is in force.
  2. Record p, sp, np and t per domain. Highlight stale p=none, any t=y and historic pct, ri or rf.
  3. List sending subdomains per client from the header_from values in reports, and confirm which record and tag covers each one (Check 2).
  4. Lock down parked domains with p=reject and an SPF record of v=spf1 -all.
  5. Compare policy_published in reports with your intended record for every domain. A mismatch means a DNS change happened outside your change process.
  6. Keep evidence. Store the dig output and one report excerpt with each change ticket.

A platform such as DMARC AI processes the aggregate reports for every client domain and uses AI to classify sending sources, so steps 3 and 5 become a filter rather than an XML exercise. How MSPs should audit client domains for DMARC risk covers the wider audit.

10

What does the NCSC say about checking DMARC?

The NCSC, the UK's National Cyber Security Centre, does not prescribe a lookup method, but its guidance makes reports the check that matters. It says reports start arriving within 24 hours of publishing, and that after moving to p=reject you should closely monitor reports for at least two weeks.

The policy of none guidance says "Within 24 hours of publishing your records you'll start receiving email reports from major email recipient domains", so no reports after a day usually means a broken rua. The two-week figure comes from the reject guidance, which still shows a percentage ramp; under RFC 9989, stage by policy level instead, optionally rehearsing with t=y. For small UK organisations, the NCSC's free Check your cyber security service includes an email security check for interception and spoofing. What the NCSC recommends for DMARC covers the full guidance.

11

FAQ

How do I check my DMARC policy without a tool?

Run dig +short TXT _dmarc.yourdomain on macOS or Linux, or Resolve-DnsName -Name _dmarc.yourdomain -Type TXT on Windows. Exactly one line starting with v=DMARC1 should come back. Read p for the policy, then sp, np and t, because they change what applies to subdomains and whether the policy is softened.

How do I test my DMARC policy with a real email?

Send a message from the domain to an external mailbox, such as a Gmail account, and open the full header with Show original. Find the receiver's Authentication-Results line and read dmarc=pass or dmarc=fail for your From domain. In Outlook, the action= value also shows which policy Microsoft applied.

Why does my subdomain show a different DMARC policy?

Because a subdomain without its own _dmarc record inherits from the Organizational Domain's record, found by the RFC 9989 tree walk. Existing subdomains get the sp value and non-existent ones the np value, falling back to p. Publish a record at the exact subdomain if it needs its own policy.

Is my domain at p=reject if the record says p=reject?

Not necessarily. Check for a second v=DMARC1 record, which cancels both, and for t=y, which makes receivers quarantine instead. Then confirm in aggregate reports that policy_published shows reject, testing is n or absent, and failing sources show disposition reject.

How long after a change can I check my DMARC policy?

Query the authoritative name server straight away to confirm the record. Public resolvers update once the previous record's TTL expires. Headers reflect the change as soon as a receiver sees the new record, and the NCSC says aggregate reports start arriving within 24 hours of publishing.

12

Final thoughts

Checking a DMARC policy is three questions: what did you publish, which record covers the domain you send from, and what did receivers actually do? The lookup answers the first, the tree walk the second (where subdomain surprises hide), and only headers and aggregate reports the third. For a portfolio, script the lookup and read policy_published and disposition from the reports. If DKIM is the part that fails, how to check if DKIM is working uses the same three-step pattern.

Related articles

Related tools

Ready to Implement?

Get authenticated mail moving in minutes: start free, book a guided demo, or talk to the team about your stack.