Read reported checks without inventing verification

SPF, DKIM and DMARC Header Results Explained

An email can contain an SPF pass, a DKIM failure, and a DMARC pass without those words describing the same test. It can also contain several reporting services with different observations. The useful task is to identify who claims to have checked what, preserve that context, and decide which question requires receiver-side evidence. A readable summary of supplied headers is not a new authentication decision.

Open the Email Header Analyzer and enter raw header text. Load example fills the input only; choose Analyze headers explicitly to parse it. Clear ends the current inspection. Start with the public example so you can learn the distinctions without introducing private correspondence.

Attach every result to its declared reporting service

An Authentication-Results field records a service identifier, followed by declared method results and supporting properties. RFC 8601 supplies the format, not a universal list of trustworthy reporting services. Read the authserv-id before reading pass or fail. In an invented field such as Authentication-Results: receiver.example; spf=pass smtp.mailfrom=sender.example, receiver.example names the claimed assessor and sender.example is attached to the SPF result. Those labels perform different jobs. Neither becomes verified because the parser recognizes its syntax. When describing the output, say that a named field reports a result, rather than saying that this website authenticated the message or contacted the named receiver.

Keep all Authentication-Results fields separate in their original source order. Two fields containing the same authserv-id remain two declarations, not one combined assessment. One field can also contain several method clauses. A summary that selects whichever pass looks most reassuring would discard precisely the context needed to investigate disagreement. Instead, identify the field, method, result token, and associated properties together. If another field disagrees, retain both. You do not yet know whether they describe different processing stages, different identities, a copied fragment, or an invented assertion. The difference is a question to investigate, not a vote in which the majority result wins.

Trust depends on the receiving environment and how it handles assertions arriving from outside its administrative boundary. A familiar hostname written into pasted text does not establish that environment. Nor can the field nearest the top automatically be declared trustworthy: a supplied block has no independently established chain of custody. The analyzer does not choose a trusted authserv-id or prove trust from position. Ask the recipient administrator which service identifiers their system writes and which inbound fields it removes or preserves. That operational answer belongs beside the source, not inside an automatic green label inferred from a domain spelling or a convenient location.

Read SPF as a reported authorization check for an identity

SPF concerns whether a connecting host is authorized to use a domain in the SMTP transaction. It is not a check of the visible display name. An Authentication-Results clause may associate SPF with smtp.mailfrom or smtp.helo; keep whichever property is actually present. Do not substitute the From address because it is easier to recognize. For a practical note, record the stated identity next to the result before discussing the sender. A business can use a separate transactional envelope domain, so different visible and envelope domains are not, by themselves, proof of impersonation or a reason to rewrite the parsed result.

Result words are not interchangeable. A reported pass asserts authorization for the checked identity; fail asserts that authorization did not hold under the reported evaluation. Softfail is a weaker negative statement, while neutral expresses no authorization judgment. None, temperror, and permerror describe different absences or evaluation problems, not three alternative spellings of malicious. A temporary error can concern the conditions of that earlier evaluation; a permanent error points toward a problem unlikely to disappear merely by retrying unchanged records. The local analyzer only parses these declarations. It makes no SPF query and cannot tell you whether the same receiver would obtain the same answer now.

Suppose a support ticket shows spf=pass smtp.mailfrom=bounces.vendor.example while the visible author uses a customer domain. The narrow finding is that the field reports SPF pass for the vendor identity. It does not establish customer-domain alignment, identify a human operator, or approve the message content. Ask which sending service and envelope configuration were intended, using administrative records outside this tool. Conversely, a declared SPF failure after a relay is not enough to diagnose the entire journey. The forwarded email authentication guide explores that separate situation; here, keep the exact method result attached to the identity it actually names.

Distinguish a DKIM signature declaration from its result

A DKIM-Signature field and a dkim=pass clause are different evidence. The signature field supplies material for a verifier; the Authentication-Results clause claims that some service evaluated a signature. Merely finding the first does not establish the second. The analyzer can describe declared DKIM tags such as d= for the signing domain, s= for the selector, and a= for the algorithm. These are declared values only. Displaying rsa-sha256, for example, is not a cryptographic check, a key-strength assessment, or a statement that the signature covers the message you received. No public key is retrieved and no signature or body hash is verified here.

Multiple signatures deserve particular care. A message can carry signatures associated with different services, and one reported evaluation can pass while another fails. Keep each result with its stated properties rather than assigning a single DKIM identity to the whole message. If the supported summary does not connect a result unambiguously to a particular signature, leave that relationship unresolved. Do not match on a familiar domain alone and manufacture certainty. A useful administrator question names the reporting service and declared selector, then asks which signature its logs evaluated. That is more actionable than reporting that DKIM was both good and bad without preserving the surrounding information.

A DKIM pass is not the same as approval by the visible author, and a failure alone does not identify the cause. Content changes, signing configuration, key availability, or other verification conditions may matter, but the pasted declaration does not let this parser select among them. Even a correctly verified signature would not certify every claim in the prose or make an included link appropriate to open. If a URL is the separate object of concern, inspect a nonsecret URL deliberately with Link Inspector; no URL is extracted and handed over automatically. Keep the signature question and the destination question separate in your investigation notes.

Keep the assessor attached

Read one declaration at a time

Inspect the public example and compare each method result with its own service identifier and properties.

Read DMARC separately from individual SPF and DKIM words

DMARC relates an authenticated identifier to the author domain rather than simply counting passes. Under the current RFC 9989 framework, an aligned authenticated identifier can support a DMARC pass; both SPF and DKIM do not have to pass together. That explains why a reported SPF failure can coexist with a reported DMARC pass when an applicable DKIM evaluation supplies alignment. It also explains why an SPF pass for an unrelated envelope domain does not settle DMARC. These are interpretation principles for receiver reports. The analyzer does not calculate DMARC alignment, discover organizational domains, retrieve policy, or independently decide which authenticated identifier qualifies for this message.

Read a declared dmarc= clause with its own supporting properties, including header.from when supplied. Do not generate a missing clause from other pass labels or from two domain strings that look similar. If no DMARC result is present, report that absence in this supplied block. Absence does not prove that the receiver skipped a check; the original view may omit fields or the copied source may be incomplete. Likewise, a parsed DMARC pass is not a current DNS-policy check. Historical receiver output and present configuration are different questions, and this offline inspection only has the text you entered to work with.

Separate authentication outcome, published handling preference, and actual disposition. A policy value such as reject is not a receipt proving that this copy was rejected. A reported failure does not tell you every local exception or handling choice, and a delivered copy does not retroactively turn that failure into pass. For troubleshooting, ask the receiver operator for the decision associated with the relevant message and processing stage. Do not ask this display to predict inbox placement, spam treatment, or future delivery. It supplies no deliverability score, spam verdict, or safety score, and a row of positive method declarations is not an endorsement of a payment request or attachment.

Investigate contradictions without flattening the evidence

Imagine one field reporting SPF pass for one identity and another reporting fail for a different identity. Calling that a contradiction before examining the properties would be premature. Now imagine identical method and identity labels with opposite results from the same declared service. That deserves a closer question, but still not an automatic finding of forgery. Preserve both fields, their ordering, and their complete raw declarations. Duplicated output, repeated processing, source alteration, or other explanations require outside evidence to distinguish. The local result should help you formulate that question precisely. It cannot determine who inserted a field or prove that an apparent reporting service ever saw the message.

ARC-Authentication-Results is not another direct Authentication-Results field to merge into the same score. ARC has separate field names and an instance structure for carrying authentication assessments through intermediaries. A declared ARC result or seal value is not proof that its chain validates, and chain validation would itself be distinct from deciding whether an intermediary deserves trust. This tool performs neither operation. Keep ARC material labelled as ARC and use it only as an additional declared context. For the different issue of transport ordering, consult Received headers and delivery delays; proximity to a trace field does not authenticate a neighboring assertion.

Parsing boundaries matter when a field contains unfamiliar syntax. The tool summarizes supported Authentication-Results clauses and preserves raw declarations; unsupported material is not silently promoted to pass or rewritten as fail. Before the first empty line, malformed header structure, orphan continuations, and non-header lines are rejected atomically. The source limit is 128 KiB of UTF-8 text, with at most 2000 lines, 500 fields, 128 Received fields, and 32 fields across Authentication-Results and ARC-Authentication-Results combined. An individual header value may contain at most 8,192 UTF-16 units after counting the retained folding whitespace and line breaks. Oversized input is rejected rather than partially assessed. The first empty line ends the header block; anything after it is ignored as body. No mail body or attachment is rendered, and header text does not become active HTML or links.

Finish with a bounded finding and the right next question

Write a finding that another operator can check: the supplied field names a particular authserv-id, reports a particular method result, and includes particular identity properties. Then state the unresolved question, such as whether that identifier is generated inside the recipient system or which signing configuration was expected. This avoids turning a formatting exercise into an unsupported accusation. For obtaining the source, use the Gmail headers guide or Outlook headers guide. If address roles are the actual dispute, continue with From, Reply-To and Return-Path instead of inferring identity from authentication words alone.

Analysis runs locally in a terminable worker inside an opaque-origin, isolated standalone PHP document, with a six-second job deadline and no analysis API or DNS access. A timeout is an incomplete local job, not an authentication failure. Clear stops work and wipes input and derived DOM; becoming hidden and the pagehide, pageshow, and freeze lifecycle events do the same. There is no storage, history, Copy command, PDF, export, or project workflow. These boundaries do not promise absolute memory erasure or protection from browser extensions, a compromised device, or a hostile parent replacing the genuine interface. Ordinary surrounding-page traffic is separate from local header parsing.

Before discussing the finding elsewhere, follow sharing email headers safely to reduce unnecessary disclosure. Redaction can also remove context, so explain which relationships remain unresolved without inventing replacement evidence. Return to the Email Header Analyzer hub for the complete inspection scope or the Tools directory for a different task. The strongest conclusion here stays deliberately specific: supplied text contains these separate declared results. It does not establish mailbox ownership, independently verified sender identity, message authenticity, or whether the recipient should act on the contents. Those decisions require evidence this header-only parser does not possess.

Questions about SPF, DKIM and DMARC Results

Does a parsed SPF, DKIM or DMARC pass mean the analyzer verified it?

No. Pass is a result token declared in the supplied header. The analyzer does not repeat an SPF query, retrieve DNS records, verify DKIM cryptography, or calculate DMARC alignment. Keep the declaring service and identity properties attached to the token when describing it.

Which Authentication-Results field should I trust when several appear?

This tool does not select one. Every field stays separate in source order, even when identifiers repeat. A trusted receiving-system configuration and source provenance are needed to assess the reporting boundary; neither a recognizable authserv-id nor its position proves that boundary from pasted text.

Why can DKIM fail while DMARC passes?

They are different evaluations. A reported aligned SPF pass can support a reported DMARC pass even if DKIM fails; another qualifying DKIM signature can also matter. Do not reconstruct that conclusion from labels alone. Read the separate DMARC declaration and preserve any ambiguity about identities or signatures.

Continue with another email-header guide

Interpret a supplied header in the Email Header Analyzer or choose a related task below. Explore more practical utilities in all free tools.

Make the next question precise

Inspect the reported evidence

Return to raw headers and keep parsed declarations distinct from checks that require receiver logs or verification.