Guides DMARC
What is DMARC?
DMARC checks the domain in the visible sender address. You choose what to ask receivers to do if the check fails and where reports should be sent.
When you send as info@example.com, the receiver needs to check that example.com is being used legitimately as the sender. Permission for the sending server or a valid signature for example.com can show this. Either check is enough. With DMARC, you also state how you want the mail handled if neither check succeeds.
One message from example.com
Someone else can put info@example.com in the sender field. The receiver therefore does not trust that address alone. It looks for additional evidence that use of example.com is authorized. There are two ways to provide it:
- The server is allowed to send for example.com. The receiver checks permission in the domain settings.
- Your mail provider signed the message for example.com. The receiver checks that digital signature.
Either check can show that example.com is being used legitimately as the sender. One successful check for example.com is therefore enough to pass DMARC. The server check is called SPF and the signature check DKIM. Even if SPF fails, DMARC can pass when DKIM succeeds for example.com.
What you ask the receiver to do
You can also state your preferred handling when neither sender check succeeds. Publish that agreement with your domain in DNS, the domain settings. The text record containing the agreement is called the DMARC record. For example.com, it is at _dmarc.example.com.
A cautious starting point is v=DMARC1; p=none; rua=mailto:dmarc@example.com. Replace the domain and use an address that can actually receive reports. The parts mean:
v=DMARC1: this is a DMARC recordp=none: request no change in handling solely because of DMARCrua=mailto:dmarc@example.com: request aggregate reports at this address
Other p values ask for more: quarantine asks the receiver to treat mail that fails DMARC as suspicious. reject asks for rejection. The receiver also uses its own policy and decides whether to reject or deliver the mail. You may not receive reports either. Do not start by choosing the strictest value.
When does a check belong to your sender domain?
For the mail from example.com, the sender address and the passing check used the same domain. That connection matters: a passing check for an unrelated domain proves nothing about example.com. In DMARC, this relationship between the domains is called alignment.
To check it in a message, you need three names. The address you see in your mail app is in the From field. The server check uses the domain of a technical return address for delivery notices. This is called MAIL FROM. The signature check uses the domain shown as d= in the signature. In the ordinary example, each is example.com.
The sending service might instead use bounce.example.com as the return-path domain. This is a subdomain under example.com. The default setting accepts such a difference when both names share the same Organizational Domain. This more permissive comparison is called relaxed, value r. DMARC determines the Organizational Domain through DNS. It is not automatically the last two parts of every name.
You can instead require exactly the same domain name. This is called strict, value s. Case makes no difference to this comparison. With From: info@example.com and a check that has already passed:
- A check for
example.com: aligns in both relaxed and strict modes - A check for
bounce.example.com: aligns in relaxed mode, not strict
If a service checks against its own domain, such as provider.example.net, that does not match example.com in either mode. An SPF pass for that other domain therefore does not count. If the same message has a valid signature for example.com, that is enough to pass DMARC. The SPF check for the provider's domain does not prevent this.
Choose the two comparisons separately: aspf= for SPF and adkim= for DKIM. Both default to relaxed. With aspf=s; adkim=r, an SPF pass for bounce.example.com does not count for From example.com, but a DKIM pass with d=mail.example.com does. Here too, only one of the two needs to succeed.
Keep relaxed unless there is a specific reason to require exact equality. Strict can be useful when parts of the same Organizational Domain are managed separately and you deliberately want to tighten that boundary. First check that each service can use exactly the required domain. Some providers offer a return address only on a subdomain. This is separate from choosing p=none, quarantine or reject.
Set up, test, then consider stricter policy
Do not make reject an automatic goal: forwarding and mailing lists can fail even for correctly configured mail. First assess which policy fits. Ask each sending service whether it can sign with an aligned domain of your own and which MAIL FROM domain it uses.
- Start with
p=noneand a working, monitoredrua=address. Publish one DMARC record at the right name - Map legitimate senders and indirect mail routes. Check real test messages and analyze reports received
- With
p=reject, do not rely solely on SPF: use valid aligned DKIM signatures - Assess the impact of
quarantinebefore consideringreject - RFC 9989 advises against
p=rejectfor domains whose users post to mailing lists. If you still want it, compare at least a month of none followed by at least a month of quarantine
In a DNS panel for example.com, the host name is usually _dmarc. Some panels append example.com automatically. Check the full name _dmarc.example.com and the saved value. Also manage the Organizational Domain deliberately if you send from a subdomain.
If a stricter policy affects legitimate mail, pause the rollout and work with your administrator to restore the previously tested policy at the right source name. Do not blindly remove a record: a stricter inherited policy could then apply. Previously cached settings do not disappear immediately. A DNS scan can check policy and its source, but does not prove real messages pass DMARC or get delivered.
Getting reports delivered
Use a mailbox that can receive reports, or follow a reporting service's onboarding instructions. A useful starting point is v=DMARC1; p=none; rua=mailto:dmarc@example.com at _dmarc.example.com, using your own domain and an address that actually exists. Agree who reads the reports and follows up on problems.
A destination outside your Organizational Domain must authorize reporting through DNS. For policy at example.com with rua=mailto:reports@collector.example.net, the authorization name is example.com._report._dmarc.collector.example.net. The destination administrator can publish TXT v=DMARC1 there, or manage suitable authorization through a wildcard. Follow the reporting service's exact instructions: this record belongs at the destination, not as extra DMARC policy in your zone.
Without required authorization, conforming receivers must not report to that address. Confirmed DNS authorization does not prove the mailbox works or reports will arrive. Reporting receivers usually send daily or periodically. Not every receiver reports. No reports therefore does not mean there was no mail or abuse. Check the address, authorization, mailbox filters and processing at your service.
Reading the reports
Aggregate reports give totals for groups of messages, not a list of their contents. They arrive as XML files, which a reporting service can make readable. They include counts, sending IP addresses, domains, authentication results and reported handling. Distinguish the individual SPF/DKIM checks from what they mean for DMARC.
Take a group of messages with From: info@example.com:
- SPF passes for
provider.example.net, but does not count for DMARC because that domain is not aligned - DKIM passes with
d=mail.example.com. Under relaxed DKIM alignment, that does count - DMARC can therefore pass even if the report shows fail for its DMARC SPF path
First identify the service behind an IP address and compare it with your sender inventory. Unknown does not automatically mean abuse: it could be a forgotten service or forwarding route. Investigate domains and errors, fix the affected service and test again before tightening policy. Reported handling can differ from your request because of the receiver's local policy.
rua contains aggregate data without message content. The optional ruf tag requests a different kind of report: failure details for individual messages, potentially including headers or content. This needs extra privacy consideration and is not required for a normal rollout. A rua reporting service also receives traffic metadata. Choose deliberately who processes it. Report delivery and completeness are not guaranteed.
Which policy applies to a subdomain?
If no usable record is found at the visible sender domain itself, DMARC searches DNS for applicable policy. The source of the record matters. Here, example.com is the Organizational Domain and its policy is at _dmarc.example.com:
p=applies to mail usingexample.comin Fromsp=can set a different policy for an existing subdomain such asnews.example.com. Without sp, p normally applies therenp=can apply to a subdomain that does not exist in DNS. Without np, the policy falls back to sp and then p
A missing mailbox does not make a subdomain non-existent in DNS. A valid record at _dmarc.news.example.com can also change the outcome: for From domain news.example.com, p from that record applies. Looking only at the record for example.com is therefore not enough.
DNS Radar shows the discovered source and applicable policy tag for the name checked. Review those before making a change. To protect a domain or subdomain that sends no mail with stricter policy, first rule out real sending services. Do not copy sp or np from an example by guesswork.
Limits and forwarded mail
DMARC can make direct abuse of your visible sender domain harder. It does not prevent deception using a lookalike domain or display name. Mail that passes every domain check can still contain harmful content.
Forwarding and mailing lists can cause correctly configured mail to fail too. A different server can make the server check fail. Changes to the message can break the signature. RFC 9989 therefore warns against relying only on SPF with p=reject: use valid signatures for an aligned domain and assess the impact through reports.
A message can have several signatures. One valid aligned signature is enough for the signature path. If neither that path nor the matching server check succeeds, the DMARC result is fail. If required DNS information cannot be retrieved, the result can remain unknown. A DNS scan without a real message cannot prove a DMARC pass or delivery.
Test mode and older receivers
Sometimes you want to test stricter policy with milder handling first. RFC 9989 provides test mode t=y for this. It is not a percentage. With p=reject; t=y, you request one level lower: quarantine. Quarantine becomes none. None is unchanged. Under that version, v=DMARC1; p=reject; t=y; rua=mailto:dmarc@example.com therefore requests quarantine on DMARC failure. Older receivers can ignore t. The receiver still decides whether to accept the mail.
Older records may contain pct= for partial enforcement. This tag was removed in RFC 9989. Receivers following that version ignore pct. Older receivers may still apply it. Do not use pct for a new rollout.
Primary sources
Frequently asked questions
What is alignment?
A passing check must belong to the domain in the visible sender address. This is called alignment. For mail from info@example.com, a check for example.com matches. The default relaxed mode also accepts names sharing the same Organizational Domain. Strict requires exactly the same domain, ignoring case. You choose this separately for the server and signature checks. One passing aligned check is enough.
Does DMARC cover subdomains?
Yes. A usable record at the subdomain can set a different policy from the Organizational Domain. Without one, DMARC searches DNS for applicable policy. Depending on the source and whether the subdomain exists, p, sp or np may apply. Check both the source name and policy tag in the scan.
Is p=none useful?
Yes, to request reports with a usable rua= address. A p=none policy requests no change in mail handling based on DMARC. Other filters and the receiver's local policy still apply. Receipt of reports is not guaranteed.
Why does legitimate mail sometimes fail DMARC?
Forwarding can cause the server check to fail. A mailing list can break the signature by changing the message. A forwarding service can sometimes pass on earlier check results. An additional technique called ARC provides this. The receiver decides whether to trust that information. Review real messages and reports before requesting rejection.
Do I need DMARC for BIMI?
Yes. A provider displaying your logo also wants you to request suspicious handling or rejection for mail that fails the domain check. Those policy values are called quarantine and reject. Reject is not a universal requirement for BIMI, the technique for such a logo. Conditions also apply to the sender and Organizational Domains. Check current provider requirements and legitimate sending services before tightening policy.