schedule 16-min read

How MSPs Can Avoid Breaking Client Email with DMARC

DMARC AI editorial team · Last updated

The DMARC rollout mistakes that break client mail, how to spot each one in aggregate reports, and the checklist and rollback plan MSPs need first.

01

How can MSPs avoid breaking client email with DMARC?

MSPs avoid breaking client email with DMARC by finding every sending source before enforcement, fixing alignment for each one, and moving one client at a time from p=none to p=quarantine to p=reject under change control. Almost every outage traces to a sender nobody inventoried, a broken SPF record, or a DNS change with no rollback plan.

An MSP (managed service provider) rarely breaks mail because it misunderstands DMARC (Domain-based Message Authentication, Reporting and Conformance). It breaks mail by running one change across many client domains with different senders, DNS hosts and owners. Below are the mistakes that cause incidents, how each shows up in aggregate (RUA) reports, and how to prevent it. Protocol background lives in the DMARC guide.

02

Key takeaways

  • Most broken rollouts come from unknown senders: scanners using Microsoft 365 Direct Send, and CRM, marketing or billing platforms that send without aligned DKIM (DomainKeys Identified Mail).
  • SPF (Sender Policy Framework) records with more than 10 DNS-querying terms, or two SPF records on one name, return permerror. That removes SPF as a path to a DMARC pass.
  • RFC 9989 removed the percentage tag. Rollout is p=none, then p=quarantine, then p=reject, optionally with t=y as a testing step at each level.
  • Treat every DMARC, SPF and DKIM change as a change request with an owner, a window, a monitoring period and a written rollback to p=none.
03

Why do DMARC rollouts break client email?

DMARC rollouts break client email because enforcement applies to every message that fails, including legitimate mail from senders the MSP never authenticated. At p=none those failures are invisible to users. At p=quarantine they land in junk, and at p=reject receivers can refuse them during the SMTP conversation.

Microsoft documents this clearly. In its DMARC configuration guide, inbound mail that fails a p=quarantine policy goes to the Junk Email folder, and mail that fails p=reject is "Rejected during SMTP (550 5.7.1)" with dmarc=fail action=oreject in the Authentication-Results header.

For an MSP, each incident costs client trust, and each one was usually visible in the aggregate reports before the policy changed.

04

Which mistakes actually break client mail during DMARC rollouts?

Nine mistakes cause most of the incidents MSPs see. Each is visible in aggregate reports or DNS before enforcement, and each has a simple preventive control. The table summarizes them; the sections below add detail.

Mistake Symptom after enforcement How to detect it Prevention
Unknown senders (scanners, Direct Send, line-of-business apps) Scans and alerts stop arriving or land in junk Source IPs in RUA reports with no matching inventory entry, SPF and DKIM failing Discovery before enforcement; move devices to SMTP AUTH or a connector
SaaS platforms without aligned DKIM (CRM, marketing, billing) Invoices, quotes and campaigns rejected DKIM passes for the vendor's domain, not the client's; SPF aligned only on the vendor domain Enable custom DKIM with d= set to the client domain
Forwarding and mailing lists Replies and list posts bounce for some recipients Failures from ISP or university forwarder IPs; DKIM sometimes still passes Rely on aligned DKIM; accept residual failures; avoid p=reject on domains whose users post to lists
SPF over 10 lookups Mail that relied on SPF fails intermittently or everywhere permerror in the SPF auth results; SPF analyzer lookup count over 10 Flatten or remove unused includes; move senders to DKIM
Duplicate SPF or DMARC records SPF permerror, or the DMARC policy ignored entirely Two v=spf1 or two v=DMARC1 answers in DNS One record per name; merge before any policy change
Subdomain surprises via sp and np Mail from billing.example.com rejected when only the parent was tested RUA rows where the header From is a subdomain Inventory subdomains; set sp and np deliberately
DNS changes without change control Unexplained failures after an "unrelated" DNS edit Pass rate drops on a date nobody can explain Change requests for every auth record; record the old value
Moving many clients at once Several clients report problems the same day Pass rates drop across the portfolio in one window Batch by risk; small waves with monitoring between
No rollback plan Hours of outage while someone finds DNS access Not detectable in reports; a process gap Written rollback to p=none, tested access, lowered TTL

How do unknown senders such as scanners and Direct Send break mail?

Unknown senders break mail because they send as the client's domain without DKIM and from IPs not listed in SPF. Scan-to-email printers and on-premises apps are the classic case. Once the policy reaches p=quarantine or p=reject, their messages fail DMARC and receivers act on them.

In Microsoft 365, Direct Send is anonymous SMTP to the tenant's MX endpoint. Microsoft's multifunction device and application guide explains that Direct Send cannot reach external recipients and that nothing adds DKIM. A scanner on a dynamic office IP will fail SPF too. Our guide to Microsoft 365 Direct Send, printers and DMARC shows how to move devices to SMTP AUTH or a relay connector first.

Detect it: in aggregate reports, look for source IPs that belong to the client's own office ranges or ISP, with both SPF and DKIM failing and the client's domain as header From.

Prevent it: run structured client discovery during onboarding. Ask about printers, CCTV systems, alarm panels, accounting packages and web forms, then match each answer to a report row.

Why do CRM, marketing and billing platforms fail DMARC?

These platforms fail DMARC because they authenticate as themselves, not as the client. They often pass SPF on their own bounce domain and sign DKIM with their own d= domain. Both pass, but neither aligns with the client's From domain, so DMARC fails.

Detect it: the report row shows dkim=pass and spf=pass in the authentication results, but the DKIM domain and SPF domain are the vendor's. The evaluated DMARC result is still fail.

Prevent it: enable custom domain authentication in each platform so DKIM signs with d=example.com or a subdomain of it. Where a platform supports a custom return path, align SPF as well. The full process is in how to handle third-party senders during DMARC projects.

How do forwarding and mailing lists cause failures?

Forwarding and mailing lists cause failures because the relaying server sends the message from its own IP while keeping the original From address. RFC 9989 Section 7.4 says SPF will likely fail "unless the RFC5321.MailFrom address is rewritten", while DKIM signatures "will generally remain valid" unless the content is changed.

Mailing lists that alter the subject or body break DKIM too. RFC 9989 notes that mail through lists with unmodified From lines "is frequently rejected due to a "p=reject" policy", and says domains hosting users who might post to lists "SHOULD NOT publish" p=reject.

Detect it: failures from IPs belonging to universities, ISPs or list servers, often with DKIM still passing. These rows are not spoofing.

Prevent it: make sure every legitimate source signs aligned DKIM, because DKIM survives plain forwarding. Accept a small residual of forwarding failures. For clients whose staff post to mailing lists, discuss whether p=quarantine is the right end state. More detail is in DMARC and email forwarding.

What happens when SPF exceeds 10 DNS lookups?

When SPF exceeds 10 DNS lookups, the receiver returns permerror and SPF cannot pass. RFC 7208 Section 4.6.4 says implementations "MUST limit the total number of those terms to 10", counting include, a, mx, ptr, exists and redirect. The ip4, ip6 and all mechanisms do not count.

Each new SaaS tool adds an include, and nested includes add more. A record like this can exceed the limit unnoticed:

example.com. TXT "v=spf1 include:spf.protection.outlook.com include:_spf.google.com include:sendgrid.net include:mail.zendesk.com include:servers.mcsv.net include:spf.example-crm.com mx a ~all"

RFC 7208 also says implementations "SHOULD limit "void lookups" to two", where a void lookup is a DNS query that returns no records.

Detect it: aggregate reports show permerror as the SPF result in the authentication results, often for every source at once. Run the record through the SPF analyzer to count lookups, including nested ones.

Prevent it: remove includes for services the client no longer uses, and move high-volume senders to aligned DKIM so SPF is not the only path. Our article on SPF too many DNS lookups covers the options.

Why do duplicate SPF or DMARC records break mail?

Duplicate records break mail because the standards treat them as errors, not as a merge. RFC 7208 Section 4.5 says that if the SPF record set "includes more than one record, check_host() produces the "permerror" result". RFC 9989 Section 4.10 says that if multiple DMARC policy records are returned, "they are all discarded".

Duplicates usually appear when a vendor wizard adds a second v=spf1 record, or a new DMARC record is published without deleting the old one.

Detect it: query the names directly before every change.

dig +short TXT example.com
dig +short TXT _dmarc.example.com
Resolve-DnsName -Name _dmarc.example.com -Type TXT

More than one v=spf1 answer or more than one v=DMARC1 answer is a defect. In reports, duplicate SPF shows as permerror.

Prevent it: one record per name, always edited in place. Our guide on merging multiple SPF records shows how.

How do sp and np cause subdomain policy surprises?

Subdomain surprises happen because the parent domain's DMARC policy covers subdomains that have no record of their own. RFC 9989 Section 4.7 defines sp as the policy for subdomains and np as the policy for non-existent subdomains. When np is absent, the sp or p policy applies.

So moving example.com to p=reject also moves billing.example.com and news.example.com to reject unless they publish their own record or sp says otherwise. Microsoft notes that the parent record "automatically covers all subdomains (including nonexistent subdomains)", but each subdomain still needs its own SPF and DKIM.

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

This record enforces on the parent, keeps existing subdomains at monitoring while you finish them, and rejects mail from names that do not exist.

Detect it: filter aggregate reports by header From and list every subdomain that appears. Any subdomain with failing volume needs work before the parent policy moves.

Prevent it: set sp and np on purpose, and write them into the change request. See DMARC for subdomains and non-existent subdomains and the np tag.

05

Why do process mistakes break mail across an MSP portfolio?

Process mistakes break mail because they turn a single-domain risk into a portfolio risk. A correct record can still cause an outage if it is published in the wrong window, for too many clients at once, or without a way back.

What goes wrong with DNS changes without change control?

Without change control, nobody knows who changed which record, when, or what it said before. A technician adds an include and pushes SPF past 10 lookups, or a web agency moves DNS hosting and drops the DKIM CNAMEs. The pass rate falls and nobody can link it to a change.

Detect it: a sudden drop in DMARC pass rate on a specific date, with no new sources in the reports, usually means a DNS edit. Compare the current records against your last saved copy.

Prevent it: treat every change to SPF, DKIM, DMARC and MX records as a change request. Record the old value, the new value, the owner and the monitoring window. Clarify who controls DNS for each client, especially where a registrar, web agency or the client's own staff can edit the zone. A template is in our DMARC change management policy for MSP portfolios.

Why is moving many clients at once risky?

Moving many clients at once is risky because problems arrive together and your team cannot triage them all. A shared cause, such as a common CRM or printer model, then hits several clients in one window and floods the service desk.

Detect it: a portfolio view of pass rates by client, compared week on week, shows correlated drops after a batch change.

Prevent it: group clients into waves by risk. Start with simple domains with few senders, then move complex ones. Microsoft gives similar advice: start "with a domain or subdomain with low mail volume and/or fewer potential email sources" and save the parent domain for last. Leave time between waves to read the reports.

What does a missing rollback plan cost?

A missing rollback plan turns a fixable failure into a long outage. When mail starts bouncing, someone has to find the DNS login, remember the old record and publish it. If credentials sit with a former employee or an agency, hours pass while legitimate mail is rejected.

06

How do I roll back a DMARC policy safely?

Roll back a DMARC policy by republishing the client's _dmarc record with p=none, keeping the rua address so reports keep flowing. It stops receivers applying quarantine or reject but does not fix the sender, so open a remediation task too.

A rollback record looks like this:

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

If the problem is limited, you can step back one level instead of all the way. Publishing p=reject; t=y asks receivers to apply quarantine, and p=quarantine; t=y asks them to apply none. RFC 9989 Section 4.7 defines t=y as a request that the validator "not apply the policy" as published.

How does DNS TTL affect a rollback?

DNS TTL (time to live) sets how long resolvers may cache a record. If the DMARC record has a long TTL, receivers can keep using the cached enforcing policy after you publish p=none. Your rollback only takes full effect once the old TTL has expired in those caches.

Check the current TTL before every change:

dig TXT _dmarc.example.com +noall +answer

The number after the name in the answer is the remaining TTL in seconds. Lower the TTL to a short value, such as 300 seconds, before the policy change. Do it at least one full old TTL in advance, because caches holding the old record will not see the new TTL until it expires. Some DNS hosts enforce a minimum TTL, so check the provider's documentation.

What should the rollback runbook include?

The rollback runbook should name who can change DNS for each client, where the credentials live, the exact previous record, and the trigger for rolling back. Agree the trigger in advance, for example confirmed rejection of a legitimate business-critical sender, so nobody debates it during an incident.

07

What should the pre-enforcement checklist cover?

The pre-enforcement checklist should prove that every legitimate source is authenticated, the DNS records are valid, and the team can reverse the change. Run it a few days before each policy move, and do not move a client until every item is confirmed.

  1. Sender inventory complete. Every source in the aggregate reports is either matched to a known service or classified as spoofing or forwarding.
  2. Aligned authentication for every legitimate source. Each one passes DKIM with the client's domain, or SPF with an aligned envelope domain.
  3. SPF valid. One v=spf1 record, 10 or fewer DNS-querying terms, no permerror in recent reports.
  4. One DMARC record. A single v=DMARC1 answer at _dmarc, with a working rua address.
  5. No percentage tag. Remove any leftover percentage tag. RFC 9989 Appendix C.5.2 lists it as removed, along with rf and ri. See why the percentage tag is going away.
  6. Subdomain policy set deliberately. sp and np chosen and documented, with subdomain traffic reviewed.
  7. Monitoring period met. RFC 9989 Section 7.4 suggests publishing p=none "for at least a month" before enforcement, then p=quarantine for an equally long period before p=reject. Google's recommended DMARC rollout suggests reviewing reports daily during monitoring.
  8. Change request approved. Owner, window, client contact and old record value recorded.
  9. Rollback ready. DNS access tested, TTL lowered, rollback record written.
  10. Client informed. Key client contacts know the date and who to call if mail goes missing.

The full sequence for each step is in how to move from DMARC monitoring to enforcement safely.

08

How should MSPs read aggregate reports during a rollout?

MSPs should read aggregate reports by source, not by message, and focus on legitimate sources that fail. Each report row gives a source IP, a message count, the header From domain, the DMARC evaluation and the raw SPF and DKIM results with their domains. That is enough to spot every mistake in the table above.

During each wave, watch for new source IPs, DKIM passing for a vendor domain while DMARC fails, SPF permerror across many rows, unfamiliar subdomains, and quarantine or reject dispositions on known legitimate sources. Aggregate reports explained covers the XML fields. DMARC AI processes aggregate reports for every client domain and uses AI to classify sending sources, which helps separate known services, forwarding and spoofing across a portfolio. Its notification emails can become Autotask tickets through Autotask Incoming Email Processing. Before each policy move, the free DMARC checker confirms the record that is actually published.

09

FAQ

How long should a client stay at p=none before enforcement?

RFC 9989 Section 7.4 suggests at least a month at p=none, then an equally long period at p=quarantine before p=reject. Google suggests a minimum of one week of daily review. Stay until every source is identified and a month-end billing cycle has appeared.

Can I still use the percentage tag to ramp up enforcement?

No. RFC 9989 removed the percentage tag because receivers rarely applied it accurately except at 0 or 100. Move from p=none to p=quarantine to p=reject. If you want a softer step, publish the next level with t=y, which asks receivers to apply the level below.

What should I do if a new sender appears mid-rollout?

Pause the next wave for that client. Identify the sender and its owner, then enable aligned DKIM or SPF for it. If it is business critical and already failing at enforcement, step back to the previous level or to p=none while you fix it. Resume once reports show it passing.

Does p=reject break forwarded email?

It can. RFC 9989 notes that SPF usually fails after forwarding, while DKIM often survives plain forwarding. Mailing lists that change messages can break both. Aligned DKIM on every legitimate source keeps most forwarded mail passing. Accept a small residual, and reconsider p=reject for domains whose users post to mailing lists.

Should every client in the portfolio move to enforcement on the same day?

No. Move clients in small waves grouped by risk, starting with simple domains that have few senders. Leave time between waves to read aggregate reports.

Related articles

Related tools

Ready to Implement?

Get authenticated mail moving in minutes: start free, book a guided demo, or talk to the team about your stack.