Separate the reporting receiver from the sending service

How to Read Google DMARC Reports

A Google DMARC aggregate report is a receiving-side observation about messages claiming a domain. It is not a list of everything your Google Workspace users sent, nor a receipt proving that Gmail put those messages in an inbox. Begin with the attachment and its declared scope, then investigate the particular alignment question it can actually answer.

Use DMARC Report Analyzer with an authorized attachment or a public example. In Files mode, choose files through Report files. Paste XML uses the XML report textarea. Load example fills only; Analyze reports starts processing explicitly.

Identify why the Google report reached your mailbox

First distinguish a report from Google from a report about mail sent using Google Workspace. A receiving organization reports observations associated with the domain being evaluated; your sending platform can be entirely different. Likewise, a Workspace organization can receive reports from several unrelated receivers. Google’s DMARC reporting guidance describes periodic XML reports delivered by email and recommends a dedicated mailbox or group. That mailbox is where your authorized collection begins, not an account this analyzer can open. A report notification arriving there is evidence of a received attachment, not proof that every receiving system sent feedback or that the attachment covers your whole sending program.

The reporting destination is requested through rua in the domain’s DMARC record. Google’s setup guidance distinguishes that DNS configuration from Google Admin console work and explains authorization when reports go to another domain. Those configuration steps belong to the people authorized to manage the relevant domains. This analyzer does not publish a record, verify a destination, subscribe a mailbox, or test whether reports can arrive. If the expected notification is absent, ask which destination was configured and who has access before treating the silence as an authentication result. A clean local display cannot diagnose a mailbox rule or a reporting-address delivery problem.

From the authorized report message, obtain the aggregate attachment rather than copying the notification body or using Gmail’s Show original view as XML. Show original concerns the message carrying the report; it is not the attachment’s aggregate records. Preserve the downloaded original in your approved evidence location when retention is necessary. Use Files for a supported XML, GZIP, or single-XML ZIP attachment, or Paste XML for one complete UTF-8 XML document. Do not join several documents into the textarea. The report packaging guide explains the local format restrictions, including why an archive with extra files cannot be treated as one report.

Read the reporting window before the Gmail story

Compare the metadata period with the operational question you brought. Begin and end are epoch seconds interpreted as a supported UTC reporting window, not your local mailbox date or the notification’s arrival time. A report arriving on Tuesday can describe an earlier period; its file-modification time is not a substitute. Both endpoints must exist and be valid, with begin no later than end. The analyzer does not invent a daily interval because the attachment looks like a familiar Google report. For a campaign launched near midnight, establish the intended UTC interval before deciding whether its traffic belongs in the displayed window. Preserve uncertainty when the available report does not span that activity.

Check the reporting organization, contact, report ID, and policy domain together. They help distinguish one attachment from another when a reporting mailbox contains several domains or repeated deliveries. These values come from the document; the analyzer does not authenticate a Google identity from a display name, contact string, or filename. It also does not recognize Google sending infrastructure from an IP address. A Google-labelled reporter and a source address in one of its rows occupy different roles. The former declares who generated feedback; the latter declares a connecting source observed in that feedback. Confusing the two can turn a receiving-side report into an unsupported claim about which service your organization used.

Policy_published is the report’s policy snapshot, not the domain’s current DNS state. Keep the supplied alignment modes and policy options attached when discussing the result. A legacy report may include pct, while a newer report can include np, testing, or discovery_method. RFC 9989 removed legacy pct, but provider examples and older attachments can still contain it. Retaining a historical value does not endorse it as a new configuration instruction. The reader supports legacy no-namespace XML and the RFC 9990 namespace, including correct prefix bindings, without certifying full XSD conformance. It does not rewrite a Google attachment into a newer vocabulary or resolve a mismatch by querying DNS.

Explain alignment without substituting a raw pass

Suppose a shop uses a third-party service for order updates. A row can report raw SPF pass for that service’s envelope domain while the visible author domain belongs to the shop. The raw result alone does not answer whether the identities aligned for DMARC. Read policy_evaluated spf separately from the SPF entry in auth_results, then inspect header_from and any supplied envelope identity. The same distinction applies to DKIM: a passing signature for a service domain is not automatically aligned with the shop. These are explanations of field roles, not fresh tests performed by the analyzer. The alignment guide follows both methods in more detail.

Google’s sender guidelines distinguish authentication requirements from domain alignment. Do not use a local aggregate result as a certificate that every Gmail requirement has been met. Here, a supported aligned SPF or DKIM pass yields derived DMARC pass. Both aligned methods must be fail to yield fail; remaining combinations are unknown. A missing or unsupported aligned value stays unknown with a warning, even if raw method results look favorable. An aligned DKIM pass can still settle derived status when SPF is unknown. Conversely, one fail with the other method unknown does not justify declaring failure. Keep the supplied values beside your explanation.

Read disposition independently. None is a reported handling value, not confirmation that an email reached the primary inbox, avoided spam filtering, or was opened. Quarantine and reject also need the report’s scope; they do not describe every attempt your shop made to contact customers. A supplied override reason can explain what the reporter declared about handling, but it does not turn failed alignment into pass. Inspect the row and its raw reason text without inventing a Google-specific scoring rule. This tool provides the same interpretation rules for supported reporters. A provider name neither grants extra credibility to the XML nor changes how pass, fail, and unknown are derived.

Begin with the reported domain roles

Inspect one Google report

Choose the relevant attachment and Analyze reports, then read reported alignment separately from raw authentication.

Use Google observations to ask a service-specific question

Start an investigation with a known workflow: order notifications, appointment reminders, support replies, or another authorized stream. Compare the reported header_from and authentication domains with the configuration that workflow was meant to use during the UTC window. That comparison requires your own inventory, not the analyzer recognizing a commercial provider from its network address. One source IP can have several rows with different identifiers or outcomes. Open the relevant row before describing an entire source as failing. The source IP investigation guide helps organize that question without geolocation, reputation checks, or automatic additions to a domain’s sending permissions.

For a recently changed third-party sender, a useful question is whether its expected custom signing domain appears in the supplied raw DKIM results, and what aligned DKIM value the reporter declared. Do not infer that a missing expected selector means the tool looked up the wrong key: it retrieves no keys. A selector here is report text, not a cryptographic test. Ask the service owner for the relevant configuration and time of change through your approved support route. Avoid pasting an entire report into an unrelated support form when a domain role and period explain the question. The raw XML viewer can confirm the selected record without creating a rendered HTML document.

Forwarding and mailing-list handling can introduce another context. A receiver may describe handling reasons or identifiers that differ from a direct sending path. Those declarations do not let this analyzer reconstruct the intermediary or verify ARC. If one customer reports a specific missing email, aggregate rows cannot identify that customer’s received copy or prove which path it took. Use Email Header Analyzer for original headers from that individual message when authorized. Do not forward a new copy and assume its outer headers preserve the original receiving experience. Keep the aggregate question about reported groups and the single-message question about that particular copy.

Compare Google reports without confusing populations

Select a report with the per-report filter before reading source totals, then compare genuinely comparable periods. The display weights each row by its count. For example, a passing row representing 80 instances and an unknown row representing 20 describe 80% passing and 20% unknown across those 100 instances, not one pass out of two equal messages. Unknown is not folded into fail to simplify the story. If the relevant denominator is zero, the percentage is unavailable. Preserve the count and reporting window with any percentage you discuss. A large change in share can reflect a different mix of reported streams as well as a configuration change.

Repeated attachments do not need to inflate the batch. Equal normalized reporter identity, report ID, policy domain, and period, with equal meaningful parsed content, identify duplicates; they are skipped with a visible count. That is different from assuming two similarly named files are identical. The same identity with different meaningful content is a conflict and aborts the full batch. Any processing error also rejects the batch atomically. Review the originals before choosing a newly scoped collection. The comparison guide explains why changing a filename or dropping an inconvenient row is not a sound way to reconcile conflicting evidence from report delivery.

Keep Google observations distinct from Microsoft reports and Yahoo reports. Overlapping or touching periods produce warnings, and combined counts remain reported message instances rather than unique emails. Separately, Google’s Postmaster Tools documentation describes dashboard-specific data boundaries, delayed updates, and possible omissions at low volume. A downloaded aggregate attachment is not a Postmaster dashboard export, and this analyzer has no connection to those dashboards. Do not force their percentages to match by removing rows. First establish whether the populations, selected domains, time windows, and measures actually ask the same question.

Close the review without leaving operational data behind

Use a small collection of whole reports appropriate to the question. Files are limited to 10, 4 MiB each, and 16 MiB source total; expanded XML has its own per-document and batch limits. Structural bounds, including nodes, depth, leaf text, and report and batch row limits, apply as well. A ZIP with another attachment, a nested archive, or encryption is not supported just because its name mentions Google. All XML must be UTF-8. The complete contract is on the DMARC Report Analyzer main page. A rejection describes supported processing, not whether Gmail regarded the underlying messages as suspicious or whether an organization should change its policy.

The genuine interface runs in an opaque-origin sandbox with no-store responses and connect-src none. A local worker has a 12-second deadline; no upload, DNS API, signature verification, or mailbox connection is involved. Clear stops work and removes file selections, pasted XML, state, and all derived results, including the raw viewer. Becoming hidden and pagehide, pageshow, and freeze trigger the same cleanup, so switching to another tab can end the review. There is no Copy, export, PDF, project, history, or persistent storage feature. Plan the observation before entering private reports, and do not depend on the interface keeping results while you inspect a different browser page.

Local cleanup does not guarantee secure memory erasure, remove clipboard content, or defend against extensions, screen capture, compromised software, or replacement of the genuine interface. Reports can expose domains, infrastructure, selectors, contacts, and diagnostic comments even without message bodies. Review any extract before sharing it under your organization’s rules; the analyzer does not redact automatically. End with a narrow statement about the selected Google-labelled report, its UTC period, weighted instances, and supported alignment declarations. Keep missing information explicit. Return to the main page for another guide or the Tools directory for another task. Reading this attachment changes neither Google Workspace settings nor the domain’s DNS policy.

Questions about Google DMARC Reports

Is a Google DMARC report a Google Workspace sent-mail log?

No. The reporter describes receiving-side observations about messages claiming a domain. The sending platform may be different. The file does not promise coverage of all organizational mail, identify every recipient, or prove inbox placement.

Should I paste Gmail Show original text into this tool?

No. Obtain the aggregate XML attachment, optionally in a supported GZIP or single-XML ZIP wrapper. Show original concerns one message and its headers. Use Email Header Analyzer for that separate task, not the aggregate XML textarea.

Why can a Google report differ from Postmaster Tools?

They are different reporting surfaces with potentially different populations and measures. Check scope and periods before comparing percentages. This analyzer reads your supplied aggregate files only; it does not retrieve dashboard data or reconcile the two systems.

Continue with another DMARC-report guide

Read an aggregate report in the DMARC Report Analyzer & XML Viewer or choose a related task below. Explore more practical utilities in all free tools.

Keep the receiving-side scope visible

Read the attachment locally

Return to the analyzer for an authorized report or public example, without connecting Google Workspace or a mailbox.