Skip to content
AdminKit

DMARC for subdomains: how does the sp= tag work?

A DMARC record on your organizational domain also covers its subdomains: sp= sets their policy, and without it they inherit p=. A subdomain needs its own _dmarc record only when it should behave differently. AdminKit's DMARC check reads sp and the generator builds it; this guide explains which record a receiver applies.

By AdminKit team, Netbase JSC · Updated

Key takeaways

  • One DMARC record at _dmarc.example.com covers example.com and every subdomain without a record of its own.
  • p= is the policy for the domain itself. sp= is the policy for its subdomains. Without sp=, subdomains get p=.
  • RFC 9989 adds np= for subdomains that do not exist in DNS, so you can reject mail from invented names without touching real ones.
  • Publish a separate _dmarc.<subdomain> record only when a subdomain needs its own policy, its own reports or its own rollout pace. Its sp= has no effect on deeper names.

Which DMARC record does a receiver use for a subdomain?

A receiver looks for a record at the exact domain in the message’s From: header first, then falls back to the organizational domain. RFC 7489, the 2015 DMARC specification, sets out the steps in §6.6.3. The receiver must query _dmarc at the From: domain. If no valid record is there, it must query the organizational domain instead, and the RFC notes that “this record can contain policy to be asserted for subdomains of the Organizational Domain.” Under RFC 7489, the organizational domain is the registered domain, usually found with a public suffix list.

RFC 9989, which obsoletes RFC 7489, keeps the order and changes how the parent is found. Its §4.10.1 lists where the record can come from, in order of preference: the author domain, the organizational domain of the author domain, and the public suffix domain. If nothing is found at the author domain, the receiver performs a “DNS Tree Walk” (§4.10), removing one label at a time and querying _dmarc at each level, with a limit of eight queries.

Then the policy is picked from the record that was found:

Record found atMail fromPolicy the receiver applies
_dmarc.example.comexample.comp=
_dmarc.example.comnews.example.com (exists, no own record)sp=, or p= if there is no sp=
_dmarc.example.comxyz123.example.com (does not exist)RFC 9989: np=, else sp=, else p=. RFC 7489: sp=, else p=
_dmarc.news.example.comnews.example.comThat record’s own p=

RFC 9989 §4.10.1 states the rule for the middle rows: when the record is that of the organizational domain and the author domain is a subdomain of it, the policy comes from sp if the author domain exists, or from np if it does not. In the absence of an applicable sp or np tag, the p policy is used for subdomains.

What exactly does sp= do?

sp= takes the same values as p=: none, quarantine or reject. RFC 7489 §6.3 defines it as the requested policy “for all subdomains”, which “applies only to subdomains of the domain queried and not to the domain itself”, and says that if it is absent, the p policy must be applied for subdomains. RFC 9989 §4.7 narrows the wording to existing subdomains of the organizational domain, because non-existent ones now have np.

Google’s DMARC setup guide describes the tag in the same terms for administrators: sp is optional and sets the policy for messages from subdomains of your primary domain, and if you don’t use it, subdomains inherit the DMARC policy set for the parent domain.

A record that monitors the main domain while rejecting spoofed subdomains looks like this:

_dmarc.example.com.  TXT  "v=DMARC1; p=none; sp=reject; rua=mailto:dmarc-reports@example.com"

The reverse is also valid. A domain that has reached p=reject but still has an older system sending from a subdomain can keep subdomains at sp=quarantine while that system is fixed.

What does np= add for subdomains that don’t exist?

Most spoofed subdomains are invented. An attacker can put invoices.example.com in a From: header whether or not that name exists in your DNS. RFC 9989 gives these names their own tag: np is the policy for non-existent subdomains, and §3.2.13 defines non-existent the DNS way, as a name whose query returns NXDOMAIN. Appendix C of RFC 9989 notes that np was imported from RFC 9091, the experimental public suffix extension, which RFC 9989 also obsoletes.

np lets you be strict where nothing legitimate can break:

_dmarc.example.com.  TXT  "v=DMARC1; p=quarantine; sp=quarantine; np=reject; rua=mailto:dmarc-reports@example.com"

Receivers that still implement RFC 7489 ignore np and fall back to sp or p, so do not rely on np alone to protect real subdomains. The dmarc.org overview lists support for subdomain policies and for non-existent subdomains among the things DMARC was designed to add over its predecessor, ADSP.

When should a subdomain get its own _dmarc record?

When it needs to behave differently from what the parent record gives it. A record at _dmarc.news.example.com is found first for mail from news.example.com, so its p=, rua=, alignment and other tags replace the parent’s for that name. Microsoft’s DMARC guide puts it this way: unlike SPF and DKIM, the DMARC record for a domain automatically covers all subdomains, including nonexistent subdomains, that don’t have their own record, and you can disrupt that inheritance by creating a DMARC record in the subdomain.

Common reasons to publish one:

  • A separate mail stream. Microsoft recommends sending through services you don’t control, such as bulk email platforms, from a subdomain like marketing.example.com rather than the main domain, so that problems with that service don’t affect the main domain’s reputation. A record on that subdomain lets it follow its own rollout.
  • A staged rollout. Microsoft’s rollout plan starts with a domain or subdomain with low mail volume or few sending sources, repeats the steps for the remaining subdomains of increasing complexity, and saves the parent domain for last. The dmarc.org FAQ describes the same pattern: start with a monitoring record for a subdomain, then repeat with other subdomains and eventually the top-level domain.
  • Separate reports. A team that runs a subdomain’s mail can receive its own aggregate reports through rua= on that subdomain’s record.
  • Different alignment. A subdomain whose senders are all under your control can use strict alignment while the parent stays relaxed.

When the work is done and the settings match, you can remove the extra records. Microsoft notes that once the settings are effectively identical for the parent domain and all subdomains, you can eliminate the subdomain records and rely on the single record in the parent domain.

Why doesn’t sp= in a subdomain record work?

Because receivers never use that record for deeper names. RFC 7489 §6.3 says that sp “will be ignored for DMARC records published on subdomains of Organizational Domains due to the effect of the DMARC policy discovery mechanism,” and RFC 9989 §4.7 keeps the same note. Under RFC 7489, mail from a.news.example.com with no record of its own makes the receiver jump straight to the organizational domain, example.com, skipping news.example.com. Under RFC 9989, the tree walk does query _dmarc.news.example.com, but §4.10.2 then selects the record found at the name with the fewest labels, again example.com, unless a record’s psd tag says otherwise.

In practice: if a.news.example.com sends mail, either publish a record at _dmarc.a.news.example.com or make sure the organizational domain’s sp= suits it. Putting sp=reject on _dmarc.news.example.com does not protect names below it.

How do subdomain policies interact with alignment?

Policy and alignment are separate questions. sp decides what happens to subdomain mail that fails DMARC; alignment decides whether it fails. Under relaxed alignment, the default, a message from news.example.com signed with DKIM d=example.com aligns, because both share the organizational domain. Under strict alignment it must be an exact match. Relaxed vs strict alignment covers that choice in detail; Google’s guide warns that strict alignment can cause messages from associated subdomains to be rejected or sent to spam.

How do you check and build subdomain records with AdminKit?

The DMARC check reads a record’s p, sp, rua, ruf, pct, adkim and aspf tags and explains them in plain language, with findings for sp, np, psd, pct and t. Two limits from its tool page matter for subdomains. The form on that page queries only the exact name you enter and has no organizational-domain fallback, so check the organizational domain to see the inherited policy, and the subdomain itself to see whether it has its own record. The Check API behind the tool goes further: it runs the RFC 9989 tree walk and applies an inherited record’s np when the subdomain does not exist (NXDOMAIN), while the form on the page does neither. How to read a DMARC check result explains each tag in the output.

The DMARC record generator builds the _dmarc TXT record in your browser from your chosen policy, optional subdomain policy (sp), report addresses, percentage and alignment settings. Leave its sp field blank and the record leaves subdomains to inherit p. np and t are not among the tags it builds, so add np= by hand if you want it. Publish the result in DNS and confirm it with the DMARC check.

A typical pair for a domain that sends from one marketing subdomain:

_dmarc.example.com.            TXT  "v=DMARC1; p=reject; sp=reject; rua=mailto:dmarc-reports@example.com"
_dmarc.marketing.example.com.  TXT  "v=DMARC1; p=quarantine; rua=mailto:dmarc-reports@example.com"

The main domain and every other subdomain are at reject; the marketing subdomain stays at quarantine until its reports are clean.

Which mistakes leave subdomains unprotected?

  • Assuming p=none on the parent is harmless for subdomains. Without sp=, every subdomain also gets none, so invented subdomains can be spoofed freely.
  • Two records at one name. RFC 7489 stops policy discovery when the set contains multiple records, so no DMARC policy applies. RFC 9989 discards them all and goes on to the parent’s record. Either way, the subdomain’s own record is not used.
  • Relying on sp= in an intermediate record. As above, it is ignored for deeper names.
  • Forgetting SPF and DKIM on the subdomain. Microsoft notes that each subdomain still requires its own SPF and DKIM setup for DMARC to pass. A strict sp= with no authentication on a sending subdomain rejects your own mail.
  • Skipping the reports. Aggregate reports show which subdomains actually send mail; rolling out DMARC with aggregate reports covers how to read them before tightening sp.

This guide is part of AdminKit’s email authentication cluster, which covers SPF, DKIM, DMARC and the records that protect a domain’s mail.

This guide is general information, not a security audit of your specific network — see our disclaimer for what AdminKit's guidance does and does not commit to.

Frequently asked questions

Do I need a DMARC record on every subdomain?

No. A receiver that finds no record at the subdomain falls back to the record of the organizational domain and applies its sp policy, or p if sp is absent. Microsoft's DMARC guide says the parent record automatically covers subdomains, including nonexistent ones, that don't have their own record. Publish a subdomain record only when that subdomain needs different settings.

What happens if sp is missing?

Subdomains get the p policy. RFC 7489 says that if sp is absent, the policy in p must be applied for subdomains, and RFC 9989 keeps that rule when np is also absent or not applicable. Google's DMARC setup guide says the same: without sp, subdomains inherit the parent domain's policy.

Does sp in a subdomain's own record do anything?

Not for deeper subdomains. Both RFC 7489 and RFC 9989 note that sp is ignored in records published on subdomains of an organizational domain, because policy discovery for a deeper name goes back to the organizational domain's record, not to the intermediate one. A subdomain record's p applies to mail using exactly that subdomain.

What is the difference between sp and np?

RFC 9989 splits subdomains in two. sp covers subdomains that exist in DNS; np covers subdomains that do not exist, meaning a DNS query for them returns NXDOMAIN. np came from the experimental RFC 9091 and is now part of DMARC. A receiver following RFC 7489 does not know np and applies sp or p instead.

Can I set p=none on the main domain and sp=reject for subdomains?

Yes, and it is a common way to start. Subdomains that send no mail have nothing to break, so rejecting mail that claims them blocks spoofing of invented names such as billing.example.com while the main domain is still being monitored. Check aggregate reports first in case a service already sends from a subdomain.

Want AdminKit to watch this for you?

Free checks tell you the state today. Protective DNS and monitoring tell you when it changes.

Huy Nguyen (David), Founder and CEO of Netbase JSC

Talk to an expert

Huy Nguyen (David), Founder & CEO, Netbase JSC

Ask whether this fix applies to your setup, or get a second pair of eyes on your DNS and mail records.