Introduction
Yes. A parked domain, one that sends no email, should publish a DMARC record with p=reject, an SPF record of v=spf1 -all and, if it receives no mail either, a null MX record. Without them, anyone can send mail that claims to come from the domain, and receivers have no instruction to block it.
Parked domains are easy to forget because nothing breaks when they are unprotected. That is exactly why attackers like them: a lookalike or old brand domain with no policy is a ready-made sender address. The UK's NCSC makes the same recommendation in its guidance on protecting parked domains. This article gives the exact record set, the edge cases and an MSP rollout plan. For the protocol background, see the DMARC guide.
Why does a domain that sends no email need DMARC?
Because DMARC protects the name, not the mail server. A receiver checks the domain in the visible From address against that domain's DMARC policy. If the domain has no policy, a spoofed message is judged on spam signals alone, and many get through. With p=reject, every message claiming to be from the domain fails DMARC, because nothing is authorised to send for it, and receivers are asked to reject it.
A parked domain is also the safest place to enforce. There is no legitimate mail to break, so there is no monitoring phase and no risk of blocking a forgotten sender. Contrast that with an active domain, where the move to enforcement needs weeks of evidence, as described in how to move from monitoring to enforcement safely.
Which records should a parked domain have?
Publish four records. Replace parked.example with the domain and the report address with your own:
parked.example. MX 0 .
parked.example. TXT "v=spf1 -all"
_dmarc.parked.example. TXT "v=DMARC1; p=reject; rua=mailto:dmarc-reports@msp.example"
*._domainkey.parked.example. TXT "v=DKIM1; p="
What each one does:
| Record | Purpose | Source |
|---|---|---|
Null MX (MX 0 .) |
States that the domain accepts no email, so senders fail fast instead of retrying | RFC 7505 |
SPF v=spf1 -all |
Authorises no server to send for the domain | RFC 7208 |
DMARC p=reject |
Asks receivers to reject anything claiming to be from the domain, and to send you reports | RFC 9989 |
| Empty DKIM wildcard | Optional: a DKIM key record with an empty p= means the key is revoked, so no selector validates |
RFC 6376 |
The first three are the essentials. Without a null MX, senders fall back to the domain's A record (see MX records and email authentication). The DKIM wildcard is optional: a spoofer cannot sign without your private key, but it makes the "no mail" intent explicit.
Do I need sp= or np= on a parked domain?
No. Under RFC 9989, a subdomain without its own record falls back to np (for subdomains that do not exist), then sp, then p. With p=reject and nothing else, every subdomain, existing or not, is covered by reject too. Add sp=reject or np=reject only if you prefer the record to state it explicitly. Non-existent subdomains and the np tag explains the fallback in detail.
Should a parked domain still have a reporting address?
Yes, if you can process the reports. Aggregate reports for a parked domain should be close to empty. Anything that appears is someone using the name, which is useful evidence for a client conversation about brand abuse. If the rua address is on a different domain from the parked domain, for example your MSP's reporting domain, that domain must publish an authorisation record under RFC 9990, Section 4: parked.example._report._dmarc.msp.example with the value v=DMARC1, or a wildcard *._report._dmarc.msp.example that covers every client domain at once.
What counts as a parked domain?
Treat a domain as parked when it sends no email of its own. That usually includes:
- defensive registrations of typos and lookalikes;
- other country-code variants of the main brand (
.co.uk,.nl,.deand so on) that only redirect; - old brand names after a rebrand or acquisition;
- campaign and product domains that host a website but never send;
- domains held for future use.
Be careful with two groups that look parked but are not:
- Domains that receive email but never send. Keep their MX records; drop only the null MX. SPF
-alland DMARCp=rejectstill apply. - Website-only domains with a contact form. Check whether the form sends notification mail using the domain in the From address. If it does, the domain is a sender and needs SPF or DKIM for that service before it can go to reject.
How should an MSP roll this out across clients?
Treat parked-domain hardening as its own short project per client. It is the quickest measurable improvement in most estates, because organisations usually own more domains than they send from:
- Inventory every registered domain, using registrar exports, not just the domains in the client's mail tenant. Ask about past brands and acquisitions.
- Confirm each domain is non-sending. Check for existing MX and SPF records, DKIM selectors in the mail platform and any aggregate reports already arriving. Ask the client to confirm in writing.
- Prepare the four records per domain and the
_report._dmarcauthorisation on your reporting domain if you use one. - Publish under change control, with the client's approval recorded, as you would for any DNS change. See DMARC change management for MSP portfolios.
- Verify with a lookup of each record, for example with the DMARC checker.
- Watch the reports. A parked domain that shows legitimate-looking traffic was not parked; investigate before assuming abuse.
- Add the domain to monitoring permanently. Registrars, DNS migrations and website rebuilds can silently remove records.
What do the NCSC and other authorities say?
The NCSC recommends v=spf1 -all and a DMARC record with p=reject for domains that do not send email, applied immediately rather than through a monitoring phase. Its wider email security guidance aims for p=reject on every domain. What the NCSC recommends for DMARC summarises the full position. Microsoft gives similar advice for the *.onmicrosoft.com domain of a Microsoft 365 tenant that does not send from it; see DMARC for Microsoft 365 clients.
Common mistakes with parked domains
- Publishing only SPF. SPF
-allalone tells receivers the sending server is not authorised, but the visible From domain is what users see. DMARC is what ties the policy to that address. - Leaving
p=none. A monitoring policy on a domain with no legitimate mail protects nothing. - Adding a
pcttag. RFC 9989 madepcthistoric. A percentage has no place on a parked domain, and current receivers ignore it. - Forgetting new registrations. Every new defensive domain needs the same records on day one. Make it part of the registration checklist.
- Unauthorised external reporting. Pointing
ruaat another domain without the_report._dmarcrecord means compliant receivers will not send reports.
FAQ
Does a parked domain need DMARC if it has no MX record?
Yes. Having no MX record does not stop someone using the domain in a From address. Spoofed mail is sent from the attacker's own servers, so only SPF and DMARC tell receivers to refuse it. Add a null MX as well to state that the domain accepts no mail.
Can p=reject on a parked domain break anything?
Only if the domain was not really parked. If a forgotten service sends with that domain, its mail will be rejected. That is why you check for MX, SPF and DKIM records, look at any existing reports and get the client's confirmation before publishing.
Should parked domains use sp=reject or np=reject?
They do not need to. Under RFC 9989, subdomains fall back from np to sp to p, so p=reject alone already covers every subdomain. Some administrators add sp=reject for clarity; it changes nothing in behaviour.
What DMARC record should I use for a parked domain?
v=DMARC1; p=reject; rua=mailto: followed by an address you monitor. Pair it with v=spf1 -all at the domain apex and, if the domain receives no mail, a null MX record of 0 ..
Do I need a DKIM record for a parked domain?
Not strictly. Nobody can create a valid DKIM signature for your domain without your private key. An empty wildcard key (v=DKIM1; p=) is an optional extra that states the domain has no valid keys.
Final thoughts
Parked domains are the cheapest DMARC win available: four records, no monitoring phase and no risk to legitimate mail. For an MSP they are also a clear, visible first deliverable in a new client engagement. Inventory the domains, confirm they are silent, publish p=reject with SPF -all and a null MX, and keep them in monitoring so they stay that way.