A DMARC audit checklist checks that every domain publishes one valid DMARC record. It also checks that SPF or DKIM aligns with the From domain, that aggregate reports arrive, and that your policy has moved past monitoring. This article sets out the checklist. Then it takes each item in turn: which standard sets it, how to check it, and what a pass looks like. One gap to watch: a record that validates but was never checked for alignment, or one that's stuck at p=none.
Contents
- The DMARC audit checklist
- Does every domain have a DMARC record, including parked ones?
- Is the DMARC record itself valid?
- Has the retired pct= tag been removed?
- Is rua= set, with reports actually arriving?
- Does SPF or DKIM alignment pass for every sender?
- Has policy moved past none, with subdomains covered?
- How to run this as a recurring review
- Start today
- Frequently asked questions
The DMARC audit checklist
A DMARC audit checks 6 things. Each gets its own section below, with the standard or guidance behind it.
- A DMARC record exists at
_dmarcon every domain, parked ones included. Subdomains are covered by that record or their own (RFC 9989 s4.5; NCSC) - The record is valid.
v=DMARC1comes first and there is only one record.p=and anysp=ornp=each hold a recognised value (RFC 9989 s4.7, s4.8, s4.10) - The record carries no
pct=tag, which RFC 9989 removed (RFC 9989 A.6, s9.3) rua=points to a monitored mailbox and aggregate reports are actually arriving (RFC 9989 s4.6, s4.7)- SPF or DKIM aligns with the From domain for every legitimate sending source (RFC 9989 s3.2.10, s5.3.5)
- Policy is at quarantine or reject, not stuck at none.
sp=andnp=are set deliberately for subdomains (RFC 9989 s4.7, s7.4)
- DMARC record exists at _dmarc on every domain, including parked ones
- Record is valid: v=DMARC1 first, one record only, valid p= tag
- No leftover pct= tag (RFC 9989 retired it)
- rua= is set and aggregate reports are actually arriving
- SPF or DKIM aligns with the From domain for every sender
- Policy at quarantine or reject, with sp= and np= set deliberately
Does every domain have a DMARC record, including parked ones?
Every domain you own should have a DMARC record, even ones that never send mail. This is NCSC's guidance for all domains, not only ones with an active mailbox.
All of your domains, including parked domains, should have DMARC records in place, regardless of whether the domain is used for email or not. Source: NCSC
The record sits at _dmarc.<domain> as a TXT record, so example.com publishes at _dmarc.example.com (RFC 9989 s4.5).
To check this, list every domain you have registered, including defensive registrations and marketing-only domains. Then check the DMARC record on each. The free DNS Lookup tool shows a domain's DMARC record when you give it the bare domain, such as example.com. No account needed.
Parked domains are often the ones skipped, since nobody is watching them for mail problems. ShieldMarc's guide to DMARC for parked domains covers the NULL MX and SPF -all pairing. Together, the two tell receivers the domain neither sends nor accepts mail.
You pass this item when every domain on the list returns a single TXT record at the exact _dmarc host. That includes domains that aren't currently sending mail.
Is the DMARC record itself valid?
A valid record starts with v=DMARC1, exact and case-sensitive, as the very first tag. If it doesn't, receivers must ignore the whole record (RFC 9989 s4.7).
Receivers read a record with no valid p=, or an invalid sp= or np= value, as p=none if it has a valid rua=. Without that, it gets no DMARC processing at all. So a mistyped sp= or np= value cancels even p=reject.
Other syntax errors elsewhere in the record fall back to a default where one exists. Otherwise, they are ignored (RFC 9989 s4.8, s4.10.1).
Two DMARC records published at the same name are both discarded, as if neither existed there (RFC 9989 s4.10, s4.10.1).
Pro Tip: Query the TXT records at _dmarc.<domain> directly with dig TXT _dmarc.<domain> and count them. A second DMARC record at the same name makes receivers discard both. Without TXT, dig asks for an A record and shows no DMARC record even when one exists.
To check validity, read the raw TXT record rather than a summarised checker view. Confirm there is exactly one record, starting v=DMARC1;. ShieldMarc's free DMARC checker validates the record against RFC 9989.
You pass this item when there is one record, the tag order is correct, and p= holds none, quarantine or reject. Any sp= or np= tag must hold one of those three values too.
Has the retired pct= tag been removed?
RFC 9989 removed the pct tag entirely and lists it as historic, alongside rf and ri (RFC 9989 A.6, s9.3).
Receivers working to the old rules still follow a pct value below 100 left in an old record. The policy then applies to only that share of failing mail. Removing the tag applies the policy to all failing mail, under both the old rules and RFC 9989.
t=y now marks a policy as testing and drops enforcement one level. It stands in for the retired pct=0 behaviour (RFC 9989 s4.7, A.6). Only receivers that implement RFC 9989 act on it. Receivers still working to the old rules ignore t= as an unknown tag and apply the policy in full. So on those older receivers, p=reject with t=y still means reject (RFC 7489 s6.3).
To check, read the raw record text for any pct= tag. ShieldMarc's free DMARC generator leaves out the pct field, so a record built with it never carries the retired tag.
You pass this item when there is no pct= tag anywhere in the record.
Is rua= set, with reports actually arriving?
To confirm this, you need two things. First, a valid rua= address in the record. Second, evidence that aggregate reports are actually landing there. The tag makes reports arrive, but it doesn't make a record valid.
There is one exception. A record with no valid p=, or an invalid sp= or np=, is read as p=none only when it has a valid rua=. Without a valid rua=, receivers that send reports must not send them at all (RFC 9989 s4.6, s4.7).
rua= holds one or more comma-separated mailto: addresses, so reports can go to more than one mailbox (RFC 9989 s4.6).
A mailbox at another organisational domain only receives reports if that report domain publishes a TXT record. That includes a mailbox at a provider, an MSP or another domain you own. The record is v=DMARC1 at <policy-domain>._report._dmarc.<report-domain>, for example example.com._report._dmarc.reports.example.net. A wildcard at *._report._dmarc.<report-domain> works too. Without that record, receivers must ignore the address (RFC 9990 s4).
Reports come only from receivers that choose to send them, and only on days they saw mail. They name only the last server that handed the mail over. Silence proves nothing: it can mean no problem just as easily as a receiver that never reports at all.
To check, open a raw aggregate report in ShieldMarc's free DMARC report viewer. Confirm it names the domain's known senders.
You pass this item when reports land at the rua= address regularly. They should cover the domain's known sending sources, with no large unexplained gap.
Does SPF or DKIM alignment pass for every sender?
DMARC passes only when SPF or DKIM aligns with the From domain. Relaxed alignment, the default, needs only the same organisational domain. Strict alignment needs an identical one (RFC 9989 s3.2.10, s4.7, s5.3.5).
Domains publishing p=reject must not rely on SPF alone. They must also apply valid DKIM signatures (RFC 9989 s7.4, s8).
To check, read the aggregate report rows for each sending source. Look at its SPF and DKIM results and identifier alignment, not just a pass or fail count. ShieldMarc's guide to understanding DMARC reports covers the fields involved. Where alignment fails, the guide on why your DMARC is failing works through the causes.
Do you send from Microsoft 365 or Google Workspace? ShieldMarc's guides to DMARC for Office 365 and DMARC for Google Workspace cover switching on DKIM signing with your own domain. That makes DKIM align with the From domain.
You pass this item when every legitimate sending source shows an aligned pass in reports. There should be no sending source you can't identify.
Has policy moved past none, with subdomains covered?
NCSC recommends starting at p=none as a monitoring phase, then iterating the policy once legitimate senders are identified.
RFC 9989 cautions against publishing p=reject if your users might post to mailing lists. If you still want p=reject, publish p=none for at least a month first. Then publish p=quarantine for an equally long period, comparing the results (RFC 9989 s7.4).
sp= covers existing subdomains and np= covers non-existent ones. Check that both are set deliberately rather than left to inherit by default. A missing np= takes the sp= value, and a missing sp= takes p= (RFC 9989 s4.7).
Receivers must not reject mail solely because a policy says p=reject. Without other analysis, they must still treat it as quarantine. Reject is a strong signal, not an absolute guarantee (RFC 9989 s5.4, s7.4).
You pass this item when policy sits at quarantine or reject, with clean reports behind that move. Your sp=/np= values should match your intent, not be left to chance. ShieldMarc's guides cover moving from none to reject and choosing between quarantine and reject.
How to run this as a recurring review
You regularly add and remove email sending services as you adopt new tools and retire old ones. So SPF and DKIM configuration drifts unless someone owns keeping it current, as NCSC's continuous improvement guidance notes.
A fixed schedule works well. Run the full checklist quarterly, for example, and review aggregate reports continuously in between. That way you catch a new or unauthorised sender sooner than the next full pass would.
MSPs auditing many client domains need a per-client, systematic pass rather than an ad hoc one. ShieldMarc's guide to DMARC for MSPs covers running that at scale.
ShieldMarc's DMARC health monitor re-reads the record every 5 to 10 minutes. It emails only when the effective p= policy changes, which catches regressions between scheduled audits.
For context, ShieldMarc's UK MSP DMARC audit for Q3 2026 found the average UK MSP still grades a C. On 16 October 2026, 76% of the 192 MSPs enforced DMARC, but only 39% were at p=reject.
Start today
Run the free checks first. The DMARC checker and DNS Lookup tool cover checklist items 1 to 3, and neither needs an account. The free Domain Trust Check shows where DMARC sits alongside SPF, certificates and the rest of the domain's Security Grade.
Open a real aggregate report in the free DMARC report viewer to work through items 4 and 5, rua= delivery and alignment.
Where the record is missing or invalid, rebuild it with the free DMARC generator.
Once the checklist passes, the next step is moving from a one-off check to ongoing monitoring and alerts. ShieldMarc's 30-day trial needs no card.
Frequently asked questions
What is a DMARC audit?
A DMARC audit checks whether a domain's DMARC record is valid under RFC 9989. It also checks whether SPF or DKIM aligns with the From domain, and whether the policy sits at none, quarantine or reject.
It is a point-in-time check against a checklist like this one. Ongoing monitoring is different: it needs continuous report review to catch new or unauthorised senders. A proper audit covers every domain and subdomain you hold, not only the ones sending mail today.
Does Cyber Essentials require a DMARC audit checklist?
Cyber Essentials does not require DMARC, SPF or DKIM. Anti-spoofing controls sit in separate NCSC guidance. So passing Cyber Essentials is not evidence that a domain is protected against spoofing. ShieldMarc's guide to Cyber Essentials and email security for suppliers sets out which NCSC guidance actually applies to UK public sector suppliers.
What tools do I need to audit DMARC?
A DNS lookup and a DMARC record checker cover the syntax items on this checklist. You need a report viewer for the alignment and reporting items. ShieldMarc's free DMARC checker, DNS Lookup and DMARC report viewer need no account. Syntax tools alone can't confirm that alignment passes in practice. That needs aggregate reports collected over time, and reports only cover the days a receiver actually saw mail.
What if the audit finds no DMARC record at all?
The fix is to publish a starting record at _dmarc.<domain> with p=none and a monitored rua= address, following NCSC's recommended first step:
v=DMARC1; p=none; rua=mailto:dmarc-reports@example.com
Parked or non-sending domains are the exception: they have no legitimate senders to find, so NCSC's parked domain guidance recommends p=reject for them straight away. ShieldMarc's guide to DMARC for parked domains covers the NULL MX and SPF -all specifics. For domains that send mail, don't jump straight to reject. Use the monitoring period to see legitimate senders first.
How often should a DMARC audit be repeated?
NCSC's continuous improvement guidance sets no fixed interval. It says to review SPF and DKIM configurations regularly, as organisations add and remove email sending services. Pair a scheduled full pass, quarterly for example, with continuous report review in between. That catches most changes as they happen. ShieldMarc's DMARC health monitor re-checks the record every 5 to 10 minutes. It emails only when the effective policy changes, which catches regressions between scheduled audits.
Sources
- RFC 9989: Domain-Based Message Authentication, Reporting, and Conformance (DMARC)
- RFC 9990: Domain-Based Message Authentication, Reporting, and Conformance (DMARC) Aggregate Reporting
- RFC 7489: Domain-based Message Authentication, Reporting, and Conformance (DMARC)
- NCSC: Implement a DMARC policy of 'none'
- NCSC: Continuous improvement (email security and anti-spoofing)
- NCSC: Protecting parked domains for the UK public sector
