What the check reads

The MX, SPF, DKIM and DMARC records of the domain, plus MTA-STS and TLS-RPT, which the result lists under “More checks”. The check follows every SPF include down to the last record and counts the DNS lookups on the way. When the name has no DMARC record of its own, it looks for one at its organizational domain, as RFC 9989 describes.

The test domains under badmail.dev each test one thing in their email records, most of them broken on purpose. Point this checker or any other at them and compare the answers.

Where the lookups go

Your browser sends each lookup over DNS over HTTPS to public DNS services that do not filter their answers: Cloudflare, Google, Quad9, Control D, DNS4EU and CZ.NIC. No lookup passes through this site. The privacy page says what each service can see.

The lookups that decide the verdict, for MX, SPF and DMARC, go to two of these services at once: Cloudflare and Google, while both answer. If one of them does not answer, the next service on the list takes its place. The check compares the two answers and uses them only when they match. When they differ, the verdict is Can’t tell, and the result shows both answers and how long to wait, because a recent change may still be spreading. When only one service answers, the verdict is also Can’t tell. A third answer never settles a disagreement.

Lookups that do not decide the verdict, such as DKIM keys, MTA-STS and TLS-RPT, go to one service, and the next one takes over if it fails. When a DKIM lookup gets no answer and no working key turns up, DKIM shows as No answer, in grey. When every lookup answers and none of the selectors tried has a key, it shows as Not found, also in grey. The lookup that tells whether a domain publishes any DKIM key at all goes to two services, and an answer that it publishes none counts only when both validate it with DNSSEC. DKIM then shows as Needs work.

If your network blocks all six services, the result says the lookups were blocked and that nothing was checked, and gives the commands that run the same lookups by hand. A check that cannot get an answer for MX, SPF or DMARC never says Protected.

Two readings of DMARC: RFC 9989 and RFC 7489

RFC 9989 became the DMARC standard in May 2026 and replaced RFC 7489. Some receivers may still follow RFC 7489, and nothing in your records shows which kind reads your email. So the check works out the policy each kind would apply to the same records, and the verdict follows the weaker of the two.

The two readings differ in six places.

  • The organizational domain. A name with no DMARC record of its own falls back on the policy of its organizational domain. RFC 9989 finds that domain by walking up the DNS tree from the name you check. RFC 7489 uses the Public Suffix List, the list of names under which domains are registered, such as com or co.uk. The two can reach different domains when the record that decides sits on a public suffix that publishes DMARC itself, such as a shared hosting platform. They can also differ without one: with a record only at mail.example.com, RFC 9989 applies it to a.mail.example.com, while RFC 7489 looks at example.com and, finding no record there, applies no DMARC. The check uses the Public Suffix List to find the domain RFC 7489 looks at. If the list does not load, the result says Can’t tell when that domain could change the verdict. It cannot when the policy is none, which leaves the domain open either way. For a name with a DMARC record of its own, the list also says which domain’s SPF a forger can use, so it counts unless aspf=s is set. Otherwise it only matters when the name you check is at reject and its subdomains are closed.
  • Names that do not exist. RFC 9989 includes np, a policy for subdomains that do not exist, first defined in the experimental RFC 9091. RFC 7489 has no np and applies sp to those names too.
  • Public suffixes. Under RFC 9989, the operator of a public suffix can mark its record with psd=y, and its policy then covers names below it that have no record of their own. RFC 7489 has no psd tag and never applies that policy.
  • More than one record. Both standards ignore a name that has more than one DMARC record. RFC 9989 then goes on up the tree. A receiver that follows RFC 7489 stops there and applies no DMARC to the email.
  • Capital letters in tag names. RFC 9989 is not consistent here: its grammar reads the tag v without regard to case, while its section on tags points to DKIM, which reads tag names exactly as written. This check takes the strict reading, so under RFC 9989 P=reject sets no policy at all, and V=DMARC1 gets a fix that writes it as v=DMARC1. RFC 7489 reads P=reject as p=reject.
  • Testing and sampling. RFC 7489 has pct, which asks receivers to apply the policy to a share of failing email and one level less to the rest. RFC 9989 removed pct and added t=y, which asks them to apply one level less to all of it while the domain tests its policy.

When the two readings give different levels, the result names the cause, says which kind of receiver applies which policy and gives the fix that brings both to the same answer.

Can the domain be spoofed?

The effective policy is the DMARC policy that covers the exact name you check: its own p tag, or, when the name inherits DMARC from its organizational domain, sp for a subdomain that exists and np for one that does not. Receivers that still follow RFC 7489 have no np and apply sp to both, so for a name that does not exist the weaker of np and sp counts. The first rule that matches gives the answer.

  1. Can’t tell

    A lookup that could change the answer got no usable answer: no DNS service answered, two services gave different answers, or only one answered. DMARC lookups count, and so do the lookup of whether the name exists when np asks for less than sp, the lookup at the top-level domain and the Public Suffix List. The Public Suffix List does not count when the policy is none, which leaves the domain open either way. For a name that has a record of its own, a DMARC lookup above it decides the policy of its subdomains and, unless aspf=s is set, which name is its organizational domain, whose SPF counts too. That lookup counts when the effective policy is reject, or when it is quarantine and aspf=s is not set. The Public Suffix List counts for such a name unless aspf=s is set, and with aspf=s only when the effective policy is reject and its subdomains are otherwise closed. None of them counts when the domain’s own SPF, through any branch receivers evaluate, authorizes the whole internet or addresses anyone can rent, and every lookup without an answer comes after the point where that SPF has decided for senders over both IPv4 and IPv6: the domain is then Spoofable whatever DMARC says. SPF lookups count only when DMARC asks for quarantine or reject. A lookup the 10-second limit cut short counts like one that got no answer. MX lookups never count. DNSSEC validation that fails on one of the records that count gives the same answer.Source: RFC 4035 §4.3

  2. Spoofable

    No DMARC policy applies to the domain for receivers that follow RFC 9989 or for those that still follow RFC 7489 (none on the name or on its organizational domain, more than one record on the same name, or a malformed v= tag), or the effective policy is none. When the two kinds of receivers read different policies, the weaker one counts.Source: RFC 9989 §4.10.1

  3. Spoofable

    A branch of the SPF tree that receivers evaluate authorizes the whole internet or addresses anyone can rent, even with DMARC at reject: +all, ip4:0.0.0.0/0, ip6:::/0, a or mx with a /0 prefix, a range wider than /16 (IPv4) or /32 (IPv6) that holds cloud addresses, or exists without macros on a name with an address.Source: RFC 7208 §5.2

  4. Partly protected

    The effective policy is quarantine, or it is reject while sp=none or np=none leaves subdomains open.Source: RFC 9989 §4.10.1

  5. Protected

    The effective policy is reject and no subdomain policy is none.Source: RFC 9989 §4.10.1

Will your real email arrive?

A second answer, apart from the verdict, about the real email the domain sends and receives. The first rule that matches gives the answer.

  1. No email

    The domain publishes a null MX and shows no sign of sending email: SPF and DKIM answered, with no SPF record that authorizes a server and no DKIM key except revoked ones.Source: RFC 7505 §3

  2. At risk

    SPF ends in permerror: more than one SPF record, more than 10 DNS lookups, more than 2 void lookups, a syntax error, or an include or redirect to a domain without SPF.Source: RFC 7208 §4.6.4

  3. At risk

    Every DKIM key found is unusable: shorter than 1024 bits, limited to sha1, malformed, with k= or h= in capitals, or in testing mode (t=y), which verifiers treat as unsigned mail.Source: RFC 8301 §3

  4. At risk

    No MX host is usable, a null MX sits next to other MX records, or a domain that sends email, or may send it, publishes a null MX.Source: RFC 7505 §3

  5. At risk

    DNSSEC validation fails for the domain’s mail records.Source: RFC 4035 §4.3

  6. At risk

    The domain sends email but has no DMARC record or no SPF record. Gmail, Yahoo and Outlook.com require both from bulk senders.Source: Google, Email sender guidelines

  7. Unknown

    None of the above, but DKIM was not found with common selectors, the domain publishes no DKIM keys, a lookup failed, or the domain probably sends no email.Source: RFC 6376 §3.6.2.1

  8. OK

    SPF is valid, at least one DKIM key is usable, DMARC is published and every lookup answered.Source: RFC 9989 §4.10.1

The forged email

The result can show a forged email from your domain. It is an example, and nothing is sent. It shows the DMARC result such an email would get, worked out from the records the check already read, what your DMARC record asks receivers to do with it, and what Gmail, Yahoo Mail, Outlook.com and Microsoft 365 document for that case, each with its source.

It describes what the record asks and what each receiver documents, and makes no forecast. RFC 9989 §5.4 puts it this way: “The final handling of any message is always a matter of local policy and is left to the discretion of the Mail Receiver.” The same section lets a receiver reject or quarantine a message that passes DMARC.

The forged email is left out when the verdict is Can’t tell, when the name does not exist and no DMARC record covers it, when the lookups were blocked and when the DMARC record could not be read, even if an open SPF record already makes the domain Spoofable.

Source: RFC 9989 §5.4, Policy Enforcement Considerations

What receivers document

The record asks receivers to reject the email

Gmail
Says it rejects it. 550 5.7.26 Unauthenticated email from domain-name is not accepted due to domain's DMARC policy.Source: Google, Gmail SMTP errors and codes, checked 18 November 2026
Yahoo Mail
Says it rejects it. Yahoo lists a message that failed authentication checks against the sending domain’s DMARC or DKIM policy among the causes of its permanent 553 and 554 errors.Source: Yahoo, SMTP Error Codes, checked 19 November 2026
Outlook.com
Says it rejects it. Microsoft says Outlook.com rejects email that fails DMARC when the record asks for quarantine or reject (announcement, July 2023).Source: Microsoft, Announcing New DMARC Policy Handling Defaults for Enhanced Email Security, checked 18 November 2026
Microsoft 365 (default settings)
Says it rejects it. 550 5.7.1, with dmarc=fail action=oreject, when the email is detected as spoofing.Source: Microsoft, Set up DMARC to validate the From address domain for cloud senders, checked 18 November 2026

The record asks receivers to quarantine the email

Gmail
Spam folder. Google describes quarantine as sending the email to the recipient’s spam folder.Source: Google, Set up DMARC, checked 18 November 2026
Yahoo Mail
Not documented. Its sender FAQ does not say what it does when the record asks for quarantine.Source: Yahoo, FAQs, checked 18 November 2026
Outlook.com
Says it rejects it. Microsoft says Outlook.com rejects email that fails DMARC when the record asks for quarantine or reject (announcement, July 2023).Source: Microsoft, Announcing New DMARC Policy Handling Defaults for Enhanced Email Security, checked 18 November 2026
Microsoft 365 (default settings)
Says it holds it. Held in quarantine or the Junk Email folder, when the email is detected as spoofing.Source: Microsoft, Recommended email and collaboration threat policy settings for cloud organizations, checked 18 November 2026

The record asks for nothing (p=none)

Gmail
Regular filtering. Google describes none as taking no action on the email.Source: Google, Set up DMARC, checked 18 November 2026
Yahoo Mail
Not documented. Its sender FAQ does not say what it does when the record asks for nothing.Source: Yahoo, FAQs, checked 18 November 2026
Outlook.com
Not documented. The announcement covers quarantine and reject only.Source: Microsoft, Announcing New DMARC Policy Handling Defaults for Enhanced Email Security, checked 18 November 2026
Microsoft 365 (default settings)
Regular filtering. No DMARC-specific action. Other filtering still applies.Source: Microsoft, Set up DMARC to validate the From address domain for cloud senders, checked 18 November 2026

The domain has no DMARC record

Gmail
Not documented. Google blocks mail from bulk senders without a DMARC record (550 5.7.40). It does not say what it does with other email from a domain without one.Source: Google, Gmail SMTP errors and codes, checked 18 November 2026
Yahoo Mail
Not documented. Yahoo requires a DMARC record from bulk senders and says their mail may go to spam or be rejected without one. It does not say what it does with other email from a domain without one.Source: Yahoo, FAQs, checked 18 November 2026
Outlook.com
Not documented. Microsoft rejects mail from domains that send 5,000 or more messages to Outlook.com without a DMARC record (550 5.7.515). It does not say what it does with other email from a domain without one.Source: Microsoft, Fix NDR error “550 5.7.515” in Outlook.com, checked 18 November 2026
Microsoft 365 (default settings)
Regular filtering. No DMARC record. Microsoft’s own check can still mark it as spoofing (compauth=fail, reason 001).Source: Microsoft, Anti-spoofing protection for cloud mailboxes, checked 18 November 2026

What the check cannot see

DKIM keys sit under names called selectors, and DNS offers no way to list them. The check tries the selectors that common mail providers use, so a key under any other selector stays out of sight. The DKIM checker lets you type your own. When the domain’s DNS is signed with DNSSEC and answers that _domainkey does not exist, the check says the domain publishes no DKIM keys.

The check reads DNS and sends no email, so it cannot tell how a given receiver treats your messages. Lookalike domains, which swap or add a letter to your name, are a separate risk that no record on your own domain can stop.