~/blog/dmarc-reports-are-the-proofPOSTED
by Shady Nathan Tawfik · November 16, 2026 · 4 min

Your DMARC Policy Is a Claim. The Reports Are the Proof.

A DMARC record is a promise nobody verifies for you. Thirty days of aggregate reports on a domain I control showed who really sends as it: about 470 deliveries from 64 IPs, a pile of forwarders, and four messages that failed both SPF and DKIM. Reading them turned a guess into a decision.

Interactive[blog-2.0]

TL;DR: A DMARC record is a promise. The aggregate reports are the only evidence that anyone keeps it. Thirty days of reports on a domain I control showed roughly 470 deliveries from 64 IPs and nine reporters, almost all of them legitimate — and four messages that failed both SPF and DKIM. Without the reports, a policy change is a guess. With them, it is a measurement.

The scenario

I wanted to move a mail domain from p=quarantine to p=reject. Everyone who has done this knows the fear: what if a legitimate sender breaks and nobody notices for a week?

The answer is not "test it carefully". The answer is in a mailbox you already have, arriving daily as XML attachments that almost nobody opens.

Every receiver that evaluates DMARC can send an aggregate report to the address in rua. Google does. Microsoft does. Yahoo does. Every gateway that filters mail does. A domain with a rua collects a complete, independent account of who sends as it.

A domain without rua collects nothing.

What a report actually contains

An aggregate report is not a mail. It is a summary: for every source IP that sent as your domain in a 24-hour window, it lists how many messages, whether SPF and DKIM passed, and what the receiver did with them.

~/code/text text
<record>
  <row>
    <source_ip>54.240.3.25</source_ip>
    <count>11</count>
    <policy_evaluated>
      <disposition>none</disposition>
      <dkim>pass</dkim><spf>pass</spf>
    </policy_evaluated>
  </row>
  <identifiers><header_from>example.tld</header_from></identifiers>
</record>

That is the entire forensic surface. No subjects, no bodies, no recipients — by design, so the reports do not become a privacy problem. What you get is exactly what you need for a policy decision: who, how many, did it authenticate, what happened.

Thirty days, four categories

I pulled the last 30 days and aggregated them. The result fell into four groups.

1. The sending service you know about. In my case the overwhelming majority: several hundred messages from the ESP's address range, dkim=pass and spf=pass, disposition none. This is the legitimate bulk. It is also the group that makes a policy change safe.

2. Forwarders and mailing lists. Around fourteen messages had dkim=pass but spf=fail. On paper that looks like a problem; in reality it is normal. A forwarder rewrites the envelope, so SPF no longer aligns, but it keeps the signature — DMARC passes on DKIM. If you only looked at SPF, you would think the domain was under attack every day.

3. The forgotten senders. Not present this month, but this is where policy changes actually break: an old marketing tool, a form on a legacy host, a single mailbox that still sends invoices. The reports are the only reliable inventory.

4. Real failures. Four messages, all from Microsoft address ranges, all with dkim=fail and spf=fail, all already disposition quarantine. Not one of them matched a known sending service.

That fourth group is the decision. Four messages in thirty days, already treated as suspicious, no legitimate sender behind them. Moving to p=reject costs nothing real and removes four messages per month from the spam folder.

TL;DR: A DMARC record is a promise. The aggregate reports are the only evidence that anyone keeps it. Thirty days of reports on a domain I control showed roughly 470 deliveries from 64 IPs and nine reporters, almost all of them legitimate - and four messages that failed both SPF and DKIM. Without reading those reports, a policy change is a guess. With them, it is a measurement.

Findings

ObservationWhat it means
~470 deliveries, 64 IPs, 9 reportersthe reports cover the whole path, not one provider
~450 dkim=pass + spf=passthe legitimate bulk is aligned and safe under reject
~14 dkim=pass + spf=failforwarders — DMARC passes on DKIM, not on SPF
4 dkim=fail + spf=failgenuine failures, already quarantined, no known sender
9 reportersone provider is not a sample; nine is a picture

The chain

rua addressDaily XML reportsParse and aggregateKnown sendersForwardersReal failuresPolicy decisionRejectInvestigate first

Why p=none is decoration

p=none produces exactly the same reports and enforces nothing. It is a monitoring mode that many organisations enable and then forget for years. The reports keep arriving, nobody reads them, and the policy stays a statement of intent.

The trap is subtle: p=none looks like DMARC in every audit checklist. The record exists, the syntax is valid, the rua is set. Only the number at the front differs — and that number is the whole product.

Detection: the weekly ten minutes

There is no dashboard needed. Four commands, once a week:

~/code/text text
fetch the last seven days of reports
aggregate by source IP and result
check for new source IPs
check for anything that failed both SPF and DKIM

Anything new in category 3 or 4 is a conversation. Everything else is noise you can ignore with confidence, because you have seen it before.

The one number that matters over time: how many messages failed both. If it stays at zero, you can go to p=reject. If it grows, something legitimate is misconfigured — or someone is testing whether you read your reports.

Closing

Mail authentication is unusual in security: the evidence arrives by mail, for free, every single day, whether you ask for it or not.

The difference between a domain that is protected and a domain that merely claims to be is not the record. It is whether someone opens the reports and turns them into a decision.

Once an attacker gains access, they can perform the following actions:

Read every sender that uses your domain: providers, gateways, forgotten marketing tools, a former agency, a single mailbox that still sends.
Identify shadow senders before an attacker does - they are the weakest path to a convincing forgery.
Spot the gap between a published policy and an enforced one: p=none reports the same data and stops nothing.
Decide a policy change on evidence instead of fear - and keep proof for the next auditor or insurer.
Shady Nathan Tawfik

Founder & Cyber IT Security Engineer in Bielefeld. Offensive security, audits and full-stack builds. More at

For educational purposes only. Every reproducible command targets my own lab environment (localhost), never a live system. © 2026