An invoice email shows your supplier’s address. A technical summary says that its sending check passed and its digital signature passed. That sounds reassuring. But what if both checks succeeded for an entirely different company’s domain?
To understand the mismatch, begin with the address the reader sees. In [email protected], the domain is the name after @, example.com. This address appears in the message’s From field, usually labelled From or Od in the mail program. A displayed name such as “Supplier Accounts” can accompany it. The computer delivering the email is a mail server: software running on a computer that transfers messages to another mail service. The visible address is part of the message that computer carries, not a guarantee of who operates the computer. 6
The delivery process also uses a return address for failures. An external invoice service may use its own domain there so it can handle failed delivery notices. That domain has a delivery job, distinct from the visible author address.
Separately, a sending service may attach a checkable digital mark using secret data called a private signing key. The corresponding public key lets the receiver check the mark without gaining the ability to create it. The mark identifies which domain’s public key to retrieve and which message material to verify.
The receiver can therefore encounter three domain names with three jobs: the visible author domain, the delivery return domain and the signing domain. Legitimate arrangements can make them different. The next step is to ask which successful check supports the visible author domain. 6
A successful check needs to be connected to the identity you are being asked to trust. Otherwise, somebody could authenticate their own domain correctly while writing your supplier’s address in the visible From field. DMARC supplies that missing connection. It also lets domain owners publish how they would like failing messages handled and request feedback. 6
A valid badge from the wrong company
Think of a visitor showing a genuine staff badge from Company B while claiming to represent Company A. The badge can be genuine without proving the particular claim. Email has a similar problem because its technical sending identities and its visible author address are not necessarily the same. 6
Two earlier checks provide the raw material. SPF checks whether a sending server may use the domain in the delivery return address. DKIM checks a signature associated with a signing domain and selected message content. DMARC asks whether a successful result belongs to a domain connected to the domain in the visible From address. That connection is called alignment. 6
Here, “authentication” means performing these technical origin checks, not signing a human into an account. A receiver first evaluates the sending permission or the message signature. It then compares the domain from a successful check with the domain the reader sees in From. A matching name without a successful check is not enough; nor is a successful check without the required match. DMARC uses the combination. 6
Take an email displaying [email protected]. The delivery return address uses mailer.example.net, and its DKIM signature also uses example.net. Even if both underlying checks pass, neither domain is aligned with example.com. Two successful checks for the wrong domain do not make a DMARC pass. 6
Now suppose the invoicing service signs the message correctly with example.com. That valid, aligned DKIM signature can satisfy DMARC even if SPF still uses the unrelated service domain. The rule is at least one successful, aligned route: SPF or DKIM. It does not require both to pass for every message. 6
Notice what changed in the second version of our example. We did not ask the receiver to ignore a failure or merely trust the service’s logo. The business arranged a valid signature associated with its own domain. That gave the receiver a checked route supporting example.com. The unrelated delivery return address could still do its legitimate job of handling delivery failures. DMARC is useful precisely because it can distinguish that working arrangement from two valid checks of an unrelated domain. 6
Alignment has a strict form, which requires identical domains, and a relaxed form, which allows related names under the same base domain, called the organizational domain, as determined by the protocol’s rules. For example, billing.example.com and example.com can align in the relaxed form. Similar-looking spelling or shared branding does not establish that relationship. example.net does not become aligned with example.com because somebody calls both services by the same business name. 6
What the domain asks the receiver to do
A domain owner publishes its DMARC instructions in DNS, the Internet directory for domain information. The receiving mail service looks up the policy applicable to the visible author domain. This is why a check should start with the domain used in the email, not automatically the address of the company’s website. 6
Three policy values are useful to recognize. p=none requests no change in message handling specifically because DMARC failed. p=quarantine asks the receiver to treat failing messages as suspicious, for example by placing them in a spam folder. p=reject asks for rejection. The receiving service still makes its own final decision; these are not commands that guarantee identical behavior across the Internet. 6
A monitoring policy is therefore not the same as no DMARC policy, but neither is it the same as asking for rejection. And a DMARC pass is not a promise of inbox placement: a receiver may still reject a message for other reasons. Keeping those distinctions prevents a simple green badge from promising too much. 6
Reports help the owner see what needs fixing
Suppose our business’s ordinary staff mail is aligned, but its new invoice service is not. If it immediately asks for rejection, legitimate invoice notices may be affected. A monitoring stage can help reveal that mismatch before stronger handling is requested. The point is to learn which real sending arrangements need correction, not merely to publish a stricter-looking setting. 6
Monitoring does not happen automatically just because the record says p=none. To request aggregate reports, the policy needs a destination marked rua. Reports also depend on receiving services that participate. Without the reporting destination, the policy does not request those reports. 6
An aggregate report is a summary of observations, such as sending sources, message counts and authentication results. It is not a copy of every invoice and not a complete record of all email on the Internet. The information can help an administrator separate a forgotten legitimate service from unexplained use of the domain. It also reveals traffic and infrastructure information, so choosing an outside reporting service means choosing who receives that information. 7
A report needs interpretation against the services the business actually uses. If an administrator sees an unfamiliar sending source, it might be an old invoice application that was missed in the setup, or it might be someone making an unauthorized domain claim. The summary alone does not establish which explanation is right. Comparing it with the business’s known sending arrangements gives the investigation a starting point. Correcting the legitimate service and observing later messages is more useful than treating every failure count as a count of criminals. 7
This is also why a monitoring stage needs an owner. Receiving summaries without looking at them does not repair the invoice service. Someone must identify the legitimate sources, make the needed changes and judge when stronger handling can be requested without unnecessarily disrupting those sources. That is a practical use of the reports, not a requirement for the business owner to read technical data every morning. 6 7
In a small business, this can be handled by the existing mail administrator or provider. The useful explanation is concrete: “The invoice service was failing because it signed with an unrelated domain; we corrected that and checked new messages.” That tells you more than “DMARC is enabled”.
What still needs a human decision
DMARC protects a particular domain claim, not every possible way of impersonating a supplier. A criminal can use a misleading display name or a different, similar-looking domain with valid authentication of its own. The domain check does not decide whether that business is the supplier you intended to contact. 6
A criminal who takes over a genuine supplier account may also send through its normal, correctly configured service. Such a message can pass domain authentication while carrying a fraudulent payment instruction. That conclusion follows from the protocol’s purpose and the real-account fraud described by the FBI. 6 12
DMARC makes a useful question answerable: does at least one successful authentication check support the visible domain? It does not answer the next question: did an authorized person approve this new bank account? An established independent confirmation route remains necessary when money is about to be redirected.