Somewhere last week, a proposal you sent for a five-figure engagement never arrived in the client's inbox. You do not know it happened. You will not know until the follow-up call gets awkward and the client says, apologetically, that they never received the email. You will apologise, resend, and never find out why.
The odds are strong that the email did not fail. It was delivered. It just landed in a spam folder because your domain is missing one DNS record.
This piece is about that record.
The invisible failure mode
Email deliverability does not fail in one loud, obvious event. It erodes.
A receiving mail server does not owe you an explanation. When it decides your message is suspicious, it does not bounce it back with a reason. It scores it against dozens of signals, subtracts a few points from your sender reputation, and drops the message into a filter folder the recipient may never open. If enough messages score the same way, your reputation drops, and the next message is more likely to be filtered too. The tighter the receiver's spam heuristics (Google Workspace and Microsoft 365 are among the strictest), the faster the erosion.
Your open rates drift down. Your reply rates drift down. Nobody calls to tell you. You assume the market is soft, or the copy is weak, or the outreach list is stale. Sometimes all of those are true. Often the real answer is a two-line DNS record you never set.
What we found scanning peer studios
We spent the last stretch of the year studying the studios we admire, from Focus Lab to Instrument to Work & Co. Part of that study was visual. Another part was infrastructure. We wrote a small internal scanner that checks the same DNS and security signals we would want a prospect to know about before we send them a first email.
The pattern we saw across the sample was the same pattern we saw when we scanned our own domain a year ago: SPF was almost always present, DKIM was hit-or-miss, DMARC was mostly missing or set to a monitor-only policy with no reporting address. These are agencies with real design reputations, real client lists, and real work to protect. None of that changes the DNS.
That is the important thing to understand. This is not a small-business problem or a technical-team problem. It is a category-wide gap, in agencies and out. Nobody is checking, and the receivers that matter have quietly been raising the bar.
Broader data backs the pattern. A February 2026 scan of 5.5 million internet domains by DmarcGuard found only 30.4% had adopted DMARC at all, and just 12.8% were enforcing a policy strict enough to actually block spoofed mail. EasyDMARC's April 2026 report on a more targeted business-domain index put valid DMARC adoption at 52.1%, up from 47.7% a year earlier. Enterprise adoption is stronger (research on the Fortune 500 puts the record-published figure north of 90%), but even there, only about six in ten enforce the strictest policy. The rest are publishing DMARC and doing nothing with it.
SPF, DKIM, and DMARC in plain English
The three records that decide most of the outcome are older than most SaaS. None of them require code. All of them live in your DNS.
SPF (Sender Policy Framework) is a public list of servers that are allowed to send email on behalf of your domain. It sits on your DNS as a TXT record that starts with v=spf1. When a receiver gets a message claiming to come from you, it checks the sending server's IP against the list. If the server is not on the list, the check fails.
The catch: most brands add every platform they have ever used to their SPF record and never remove the old ones. The spec allows a maximum of ten DNS lookups per SPF evaluation. Overshoot and the record silently breaks.
DKIM (DomainKeys Identified Mail) is a cryptographic signature. Your sending platform holds a private key. The matching public key sits in your DNS. Every outbound message gets signed with the private key, and receivers use the public key to verify that the message was not altered in transit and did come from a server holding your key.
The catch: DKIM keys are per-platform. If you send from a marketing platform, a CRM, and a transactional service, you have three DKIM records to keep alive.
DMARC is the policy that binds the other two together. It tells receivers what to do when a message fails SPF or DKIM: nothing (monitor), send to spam (quarantine), or reject outright. It also gives you an email address to receive aggregate reports about who is sending messages that claim to be you. Those reports are how you find out that a hosting platform you cancelled two years ago is still trying to send email from your domain, or that a spammer is spoofing your address to your own customers.
A minimal DMARC record looks like this, published as a TXT record at _dmarc.yourdomain.com:
v=DMARC1; p=none; rua=mailto:dmarc-reports@yourdomain.com
That says: enforce nothing yet, and send me the aggregate reports so I can see what is happening. It is the correct place to start. It is not the place to stop.
What actually happens when DMARC fails or rejects
Two kinds of DMARC events start showing up the moment you publish the record. Neither is loud, which is what makes DMARC hard to reason about at first.
Rejected messages. When your policy is p=reject and a message claiming to be from you fails both SPF and DKIM, the receiving mail server drops it before delivery. The recipient sees nothing. The sender may or may not get a bounce, depending on the receiver's configuration. This is the intended behavior for spoofing attempts and the reason DMARC exists at all. It is also the reason a rushed rollout is dangerous: a misconfigured legitimate sender (a marketing platform whose SPF include-line went stale, an ops tool the founder set up two years ago and forgot) gets treated the same as a spammer, and its mail silently disappears.
Aggregate reports. If you set up DMARC and are suddenly getting daily XML emails from Gmail, Yahoo, and Microsoft, those are the reports you asked for. Each one is a summary from a specific receiver, listing every source that tried to send messages using your domain over the last 24 hours: source IP, count, whether SPF and DKIM passed, whether they aligned with your From-domain, and whether the message was delivered or filtered. Raw XML is unreadable on purpose. Postmark's free DMARC monitor and Valimail's free tier both parse them into a dashboard that shows the same data in a way a human can act on.
Together these two feedback loops are how you actually maintain email authentication over time. Publish, wait for reports, fix what breaks, tighten policy.
The right rollout, not the fast one
The reason most brands either skip DMARC or leave it stuck on p=none forever is the same. Moving from monitor to enforce, without an audit, breaks legitimate email. Every service you use that sends on your behalf has to pass either SPF or DKIM (ideally both). Miss one, and the day you flip your policy to quarantine or reject, the marketing team's newsletter starts landing in spam, your invoicing platform stops delivering receipts, and the founder's personal address suddenly can't cc a client.
The safe rollout is boring on purpose.
- Publish DMARC in monitor mode (
p=none) with aruareporting address. - Collect two to four weeks of aggregate reports. Parsers like Postmark's free DMARC monitor or Valimail's free tier will turn the XML into a readable dashboard.
- Audit the senders that show up. Every legitimate sender needs SPF alignment, DKIM alignment, or both. Every unknown sender is either an old service still connected or an attempted spoof.
- Move to
p=quarantineonce every legitimate sender is passing. Keep monitoring. - Move to
p=rejectonce you are confident. This is the strict policy Google and Microsoft prefer to see from serious senders.
That whole process is a couple of hours of focused work spread across a month of monitoring. It is not glamorous. It is the difference between a pipeline that lands and a pipeline that drifts.
Why this matters more in 2026 than it did in 2022
The receivers moved the goalposts. In early 2024, Google and Yahoo both started requiring proper authentication (SPF and DKIM aligned, plus a valid DMARC record) for any sender pushing more than 5,000 messages per day to their users. That was the loud change. The quiet change since then is that both receivers have been tightening the same signals for smaller senders too. Microsoft 365 has been on the same trajectory, driven by enterprise IT teams demanding stronger anti-spoofing defaults.
The result is that the deliverability floor has risen for everyone. A brand that shipped fine in 2022 with only SPF configured is now sitting at a lower inbox rate than it was two years ago, and nobody in the building has noticed. There is no dashboard for the emails that never arrived.
How to check yours today
Five minutes, three lookups, one honest answer.
dig TXT yourdomain.com | grep spf
dig TXT _dmarc.yourdomain.com
dig TXT default._domainkey.yourdomain.com
The first returns your SPF record if one is published. The second returns your DMARC record. The third probes the most common DKIM selector; if your sending platform uses a different name, you will need the selector from that platform's dashboard.
If any of the three come back empty, you have a real gap. If your DMARC returns p=none with no rua address, you have a stub, not a policy. If SPF ends in ~all (soft fail) or ?all (neutral), you are asking receivers to guess.
Prefer a click over a command line? MXToolbox and DNSChecker both offer free lookup tools that return the same answers in a readable page.
Where we sit on this at Sapere Digital
We treat email authentication as part of the launch checklist, not an afterthought. Every site we ship for a client goes live with SPF, DKIM, and DMARC configured against their real sending stack, aggregate reports pointed at a real address, and a rollout plan documented for the client's IT lead or ops team to follow. It costs no more than an hour of work at launch and it protects the pipeline the site was built to feed.
For our own outbound, we built ourselves a small scanner that runs the same set of checks (DNS, TLS, security headers, and a set of other public signals) on any domain we are about to reach out to. If we cannot deliver a proposal to a prospect reliably, nothing else about the pitch matters. If they cannot deliver a proposal to their own clients reliably, that is a conversation worth having.
Silent failure is the worst kind. This one is fixable in an afternoon.
Sources
- Google: Email sender guidelines
- Yahoo: Sender best practices
- Microsoft: How Microsoft 365 uses SPF, DKIM, and DMARC
- RFC 7489: Domain-based Message Authentication, Reporting, and Conformance (DMARC)
- RFC 7208: Sender Policy Framework (SPF)
- Postmark: Free DMARC monitoring service
- Valimail: DMARC monitoring for free (Monitor tier)
- MXToolbox: DMARC record lookup
- DNSChecker: DMARC record lookup
- DmarcGuard: DMARC Adoption 2026 (5.5M domains scanned)
- EasyDMARC: 2026 DMARC Adoption Report (52.1%)



