DMARC record step by step: from p=none to p=reject.
A DMARC record is a DNS TXT entry under the name _dmarc prefixed to the domain, in which the domain owner sets out the policy for messages that fail DMARC and the address for reports. It is rolled out gradually: first p=none with aggregate reports, then, once the reports show that all of the domain's genuine mail passes DMARC, p=quarantine, and finally p=reject.
- Published
- Updated
- Reading time
- 9 min
- Category
- Deliverability
- Author
- DYNTRO team
Where do you add a DMARC record?
A DMARC record is added in the DNS zone of the domain the mail is sent from, as a TXT record under the name _dmarc. For the domain example.com, the full name is _dmarc.example.com. The record is created with the domain registrar or DNS provider, wherever the SPF and DKIM records already are. For a company's own domains, Microsoft 365 has no separate panel for this.
A domain's record covers its subdomains that have no record of their own, including non-existent ones. A different registered domain, however, needs its own record. This also applies to a separate domain for cold email: the record on the company domain does not cover it.
DMARC uses the results of SPF and DKIM, so both mechanisms must be working before the DMARC record is published. We explain how to set them up in our article on SPF, DKIM and DMARC.
What does a DMARC record consist of?
Since May 2026, DMARC has been described in RFC 9989, which replaced RFC 7489 and RFC 9091. Aggregate reports are described in a separate RFC 9990, and failure reports on individual messages in RFC 9991. The record is a list of tags in the form tag=value, separated by semicolons. The first must be v=DMARC1, otherwise the receiver will ignore the entire record. Unknown tags are skipped by the receiver.
| Tag | What it means | Default value |
|---|---|---|
| v | version, the only value is DMARC1 | required |
| p | domain policy: none, quarantine or reject | none if the tag is missing |
| sp | policy for existing subdomains | same as p |
| np | policy for non-existent subdomains | same as sp, or as p if there is no sp |
| rua | addresses for aggregate reports | no reports |
| ruf | addresses for failure reports on individual messages | no reports |
| adkim | DKIM alignment: r relaxed or s strict | r |
| aspf | SPF alignment: r relaxed or s strict | r |
| t | test mode: y or n | n |
| psd | whether the domain is a public suffix, e.g. a top-level domain | u, i.e. not specified |
Compared with RFC 7489, the np, psd and t tags have been added, while pct, rf and ri have been retired. The value psd=y is set by operators of public suffixes, e.g. top-level domains. The owner of a company domain can leave this tag out.
Older guides still show pct, the percentage of messages covered by the policy. RFC 9989 removed it because receivers applied it inaccurately when the value was anything other than 0 or 100. Its role has been taken over by t. With t=y, the domain owner asks receivers to apply a policy one level milder: quarantine instead of reject, none instead of quarantine. The tag does not affect reports.
How does DMARC check alignment with SPF and DKIM?
A message passes DMARC when at least one result is a pass and is aligned with the domain in the “From” field: SPF for the domain in the envelope sender address, or DKIM for the domain in the signature (d=). In relaxed mode, the same organisational domain is enough; in strict mode, the domains must be identical. RFC 9989 notes that relaxed mode is sufficient for almost all domain owners.
| “From” field | SPF or DKIM domain | Relaxed mode (r) | Strict mode (s) |
|---|---|---|---|
| anna@example.com | example.com | passes | passes |
| anna@example.com | mail.example.com | passes | fails |
| anna@example.com | example.org | fails | fails |
Before moving to reject, DKIM becomes more important. RFC 9989 requires that a domain with a reject policy does not base DMARC on SPF alone and signs its messages with DKIM. The reason: when another server forwards a message, SPF usually stops passing, while the DKIM signature usually remains valid as long as the server does not change the signed parts of the message.
Where do you start a DMARC rollout?
RFC 9989 describes the order of steps for the domain owner:
- SPF and DKIM set up so that they produce a result aligned with the domain in the “From” field.
- A mailbox for aggregate reports and a person who reads them.
- A record with
p=noneand the report address inrua. The RFC calls this monitoring mode. - Analysis of the reports and fixing the mail sources that do not pass DMARC.
- A decision on a quarantine or reject policy.
Example record to start with:
Name: _dmarc.example.com
Type: TXT
Value: v=DMARC1; p=none; rua=mailto:dmarc-reports@example.com
A p=none policy blocks nothing. RFC 9989 states plainly that the receiver must not change how it handles mail because of it. The domain owner receives reports, and messages are treated as if there were no DMARC.
Since 1 February 2024, Gmail has required a DMARC record from senders who send close to 5,000 messages a day or more to personal Gmail accounts. A none policy is sufficient, but the domain in the “From” field must be aligned with the SPF or DKIM domain. Google also recommends setting up SPF, DKIM and DMARC for your domains regardless of this threshold.
How do you read DMARC aggregate reports?
An aggregate report is an XML file that the receiving server sends to the address in rua. Microsoft says it usually arrives once a day with results from the previous day, often compressed with GZIP. Each report covers one domain and comes from one receiver. Among other things, it contains:
- the IP addresses of the servers that sent mail using the domain, and the number of messages from each;
- the SPF and DKIM results and whether each of them was aligned;
- the policy from the record and what the receiver did with the messages.
The most telling part is the difference between the authentication result and alignment. Microsoft draws attention to the pattern where SPF shows “pass” but is not aligned: the server is allowed to send, but it uses a different domain in the envelope. A typical case is a service that sends on the company's behalf and signs mail with its own domain.
In the reports, look for:
- known mail sources that do not pass DMARC, because they need to be fixed before the policy is tightened;
- unknown IP addresses with a large number of messages, because this may be someone spoofing the domain;
- receivers that do not apply the published policy.
XML files are hard to read by hand. RFC 9989 recommends processing them automatically, with your own tool or in a service that collects reports and presents them in tables.
When should you move from p=none to quarantine and reject?
When the reports show that all of the domain's genuine mail passes DMARC. RFC 9989 requires every legitimate mail source without alignment to be fixed before a quarantine or reject policy is switched on. It also notes that with infrequent sending, collecting reliable data can take many months. Microsoft recommends moving gradually, from none through quarantine to reject, and watching the reports at every step.
The example of Microsoft 365, when set to honour DMARC policies, shows what the policy changes:
| Sender policy | What Microsoft 365 does with a message that failed DMARC |
|---|---|
| p=none | nothing because of DMARC; the usual filters apply |
| p=quarantine | delivers it to Junk Email, or with a different setting to quarantine |
| p=reject | rejects it during the connection, code 5.7.509 (on its DMARC page Microsoft gives 550 5.7.1) |
The final decision rests with the receiver. RFC 9989 requires receivers not to reject mail solely because of a reject policy and, without additional knowledge and analysis, to treat it as quarantine.
For domains whose users also post to mailing lists, RFC 9989 recommends at least a month with p=none and the same with p=quarantine before reject, comparing the results. A campaign domain usually has no such users, but the principle holds: each stage should last long enough for the reports to cover a full sending cycle.
Example company, domain example.com, report address dmarc-reports@example.com in every record.
| Stage | Record value without rua | When to move on |
|---|---|---|
| monitoring | v=DMARC1; p=none; np=reject | all known mail sources pass DMARC with alignment |
| quarantine | v=DMARC1; p=quarantine | reports show no genuine mail being sent to spam |
| reject test | v=DMARC1; p=reject; t=y | a full sending cycle without new errors |
| reject | v=DMARC1; p=reject | keep reading the reports every week |
From the start, the np=reject tag applies only to subdomains that do not exist in DNS, so it should not affect genuine mail.
Who should read DMARC reports?
Microsoft recommends sending reports to a dedicated shared mailbox or Microsoft 365 group rather than a user's mailbox, and reviewing them weekly at the start of the rollout and monthly after that. A good choice is a role-based mailbox, e.g. dmarc-reports@, with an owner who knows what to do with the results and a stand-in for holidays.
The report address can be in a different domain, e.g. at a service that analyses the reports. That domain must then confirm its consent with a TXT record named example.com._report._dmarc in its own zone; otherwise receivers will not send reports there. RFC 9990 describes this safeguard so that nobody can flood someone else's mailbox with reports.
Failure reports on individual messages (ruf) are optional. Microsoft 365 does not send them at all, even when the record has this tag. Reports contain the IP addresses and domains of the servers that send mail on the company's behalf, so it is worth controlling access to them in the same way as access to other technical data.
Summary.
- A DMARC record is a TXT record under
_dmarcprefixed to the domain. The first tag isv=DMARC1, followed by the policypand the report addressrua. - RFC 9989 of May 2026 added the np, psd and t tags to the record and removed pct.
- To start,
p=nonewith reports. This is measurement, not protection for the domain. - Before quarantine and reject, every genuine mail source that does not pass DMARC has to be fixed, and with reject, mail has to be signed with DKIM.
- Reports go to a role-based mailbox that has an owner.
We run email and LinkedIn campaigns to decision-makers, from data and domains to replies, while your salespeople focus only on conversations. See the service: B2B email campaigns.
Book a callQuestions and answers.
Is p=none enough for a campaign?
To meet Gmail's requirements, yes: since 1 February 2024, senders who send close to 5,000 messages a day or more to personal Gmail accounts must have a DMARC record, and the policy can be set to none. As protection for the domain, p=none is not enough. According to RFC 9989, such a policy must not change how messages are handled, so mail that spoofs the domain reaches recipients just as it did before. A p=none policy is a measurement stage, not the goal.
Where do DMARC reports go?
To the addresses in the rua tag, usually once a day, as an XML file that is often compressed with GZIP. Without an rua tag, receiving servers do not send aggregate reports. An address in a different domain from the one with the record only works if that other domain confirms its consent with a record under the name _report._dmarc. It is best to send reports to a role-based mailbox or to a service that processes them, and to appoint a person who reads them.
Sources.
- RFC 9989, Domain-Based Message Authentication, Reporting, and Conformance (DMARC)datatracker.ietf.org
- RFC 9990, DMARC Aggregate Reportingdatatracker.ietf.org
- RFC 9991, DMARC Failure Reportingdatatracker.ietf.org
- RFC 7960, Interoperability Issues between DMARC and Indirect Email Flowsdatatracker.ietf.org
- Microsoft Learn, Set up DMARC to validate email in Microsoft 365learn.microsoft.com
- Microsoft Learn, Email nondelivery reports (NDRs) and SMTP errors in Exchange Onlinelearn.microsoft.com
- Google, Email sender guidelinessupport.google.com
