Give the report its receiving-side context
How to Read Microsoft DMARC Reports
Microsoft DMARC reporting can involve several roles: Microsoft 365 as a receiving system, your organization as a sending domain, and Outlook as the application displaying a report email. Establish which role you mean before interpreting the attachment. Aggregate XML describes reported groups of message instances, not a searchable log of everything Outlook users sent or received.
Open DMARC Report Analyzer and use Files to choose files through Report files. For one complete XML document, use Paste XML and the XML report textarea. Load example fills only; choose Analyze reports explicitly. No Microsoft account connection is required.
Distinguish a Microsoft reporter from an Outlook mailbox
A report delivered to an Outlook mailbox need not have been generated by Microsoft. Conversely, a report generated by a Microsoft receiving system can describe messages sent through another organization’s infrastructure. The application displaying the notification is not the reporter identity. Begin with your authorized report collection, then inspect the organization, contact, report ID, policy domain, and period declared inside each attachment. A subject or recognizable display name helps locate a candidate file, but the analyzer does not authenticate it. Keep this distinction when someone asks for the Microsoft report: clarify whether they mean feedback generated by Microsoft or feedback about their Microsoft-hosted sending domain.
Microsoft’s DMARC documentation states that Microsoft 365 sends aggregate reports for valid rua=mailto destinations when the receiving Microsoft 365 domain’s MX points directly to Microsoft 365. It also states that Microsoft 365 does not send DMARC forensic reports. That is a specific service condition, not a promise from this analyzer that all Microsoft-related traffic will appear. Do not apply it indiscriminately to every product called Outlook or to every organization using the Outlook client. The distinction between aggregate feedback and per-message forensic material also matters when deciding whether an attachment belongs in an XML aggregate reader at all.
Agree who is responsible for collecting feedback and which domain is being reviewed. A team report mailbox can contain several sending domains, old policy periods, and duplicate deliveries. Keep the collection organized under your existing access and retention rules instead of forwarding every notification to colleagues. This analyzer has no mailbox subscription, Microsoft Graph integration, automatic report retrieval, account connection, or monitoring schedule. If the organization needs an ongoing collection process, that is separate from opening a few authorized attachments locally. Decide the immediate question first, such as whether an expected signing identity appears during a known period, so the review does not become an unbounded tour of unrelated operational records.
Move from the notification to the actual report data
Locate the aggregate attachment rather than pasting the text Outlook displays around it. Microsoft describes aggregate reports as typically daily email attachments containing XML, often GZIP compressed. A notification subject can distinguish a report domain from a submitter, but it is not a replacement for metadata inside the report. Do not paste internet message headers, a transport error, or a message-trace table into the XML report textarea. Those are different evidence types. Preserve the original attachment through an approved process when it must be retained. The analyzer can read supported packaging locally; it does not need the mailbox password or permission to search the rest of the folder.
Files accepts .xml, .gz, and .zip within the documented resource bounds. Paste XML accepts one UTF-8 XML document, not an .eml message, an Outlook item, base64 attachment text, or multiple reports concatenated together. A ZIP must contain one XML file and optional directory entries only. Extra files, encrypted entries, nested archives, ambiguous structure, and integrity failures are rejected. A convenient ZIP downloaded from another system may therefore be outside this tool’s contract even if one of its members is useful XML. The XML and archive guide explains the difference between collecting a whole supported report and editing packaging or content to conceal a rejection.
After analysis, read the actual UTC begin and end before comparing the attachment with an incident. They are reporting-window epoch values, not when Outlook synchronized the email or when a user downloaded the file. Required metadata must be present and valid; an absent date is not reconstructed from a filename. Legacy no-namespace reports and the RFC 9990 namespace are supported, with correctly bound prefixes. Unsupported extension namespaces produce warnings and are not interpreted as standard fields merely because element names resemble them. This bounded structural reader is not full XSD certification. Successful parsing neither confirms Microsoft generated the attachment nor proves it contains every expected observation for the declared domain.
Read alignment separately from Microsoft handling signals
A project-management service can send notifications with your visible author domain while using its own envelope or signing domain. Raw SPF or DKIM pass in auth_results does not establish DMARC alignment with your header_from domain. Read the SPF and DKIM values under policy_evaluated for the reporter’s aligned assessments. Then inspect supplied identifiers, signing domains, selectors, and raw results to understand the question those values raise. The analyzer does not rerun SPF, retrieve DNS records, verify DKIM signatures, or decide alignment by recognizing the service name. Use the alignment guide when colleagues use authenticated and aligned as though they were interchangeable descriptions of a row.
Derived DMARC passes when either supported aligned method is pass and fails only when both are fail. Other combinations remain unknown. For example, an aligned SPF fail together with an absent DKIM alignment value is unknown, even if one raw DKIM entry says pass. The missing value could change the conclusion and cannot be supplied by a favorable-looking domain. Missing or unsupported aligned fields generate warnings rather than automatic failure. An aligned pass from either method still establishes derived pass when the other is unknown. This is the same rule applied to other reporters, not a Microsoft-specific grade. Preserve the underlying values so another reviewer can follow how the displayed status was obtained.
Disposition is a separate reported handling declaration. None, pass, quarantine, reject, and other are not substitutes for the aligned method fields. Microsoft’s email authentication operations guide also distinguishes composite authentication, or compauth, as a Microsoft signal in message-header investigations. This aggregate analyzer does not calculate compauth, apply a tenant’s anti-phishing policy, or recreate Defender decisions. A reported override reason can provide context without proving the original message reached an inbox or was safe. Do not translate a quarantine disposition into a claim that the analyzer checked the tenant’s quarantine, and do not translate none into proof that all other filtering accepted the message.
Keep alignment and handling separate
Follow the Microsoft report rows
Analyze an authorized aggregate attachment locally, then inspect the reported identities and aligned method values.
Route a source question to the owner who can answer it
Use the per-report filter to isolate the relevant observation, then inspect an IP summary and its individual rows. The same address can represent several combinations of header domain, authentication identities, and results. A valid IPv4 or IPv6 value is not a verified Microsoft service address, geographic location, sender identity, or reputation rating. There is no mail-provider recognition from IP, reverse-DNS request, ASN lookup, or geolocation here. Compare the declared source with your organization’s authorized inventory through a separate support process. The failure and source investigation guide helps keep the question focused on the reported evidence rather than an assumed network owner.
For a known business workflow, identify the person who owns its sending configuration. Ask which envelope and signing domains were expected at the reported time, whether a custom domain was enabled, and whether the workflow changed recently. The XML can show declared domains and selectors but not who controlled the sending service or whether a configuration change was deployed everywhere. Do not add an unfamiliar source to SPF or relax an enforcement policy simply because a row carries a familiar domain. A report is a prompt to reconcile observations with authorized records. The raw XML viewer can help check the exact selected row as plain text, without treating a contact address or diagnostic URL as an active instruction.
If forwarding, a gateway, or a mailing list is part of the known path, give that context to the responsible administrator. The analyzer does not infer those services from an IP, verify ARC, or choose a tenant exception. A supplied policy reason remains the reporter’s declaration. For one disputed message, collect that message’s original headers with authorization and use Email Header Analyzer for the separate single-message reading. Aggregate XML cannot identify a person’s inbox item, recover its body, or reproduce message-trace events. Keep the report period and grouped count distinct from evidence about a particular received copy, even when the same business workflow appears in both investigations.
Compare Microsoft observations with a stated denominator
Read totals as weighted reported message instances. If one passing row has count 120 and one failing row has count 30, the passing share is 80% of 150 instances, not 50% because two rows are visible. An unknown row contributes its stated count to the unknown category rather than silently becoming fail. A zero denominator yields an unavailable percentage. Counts are not numbers of staff members, recipients, or distinct emails sent from Exchange Online. When a small problematic stream disappears beneath a large passing one, inspect its row and report rather than concluding it does not matter. Conversely, a small failing denominator does not establish an organization-wide failure rate.
Repeated report deliveries are handled by normalized reporter identity, report ID, policy domain, begin, and end, together with meaningful parsed content. Equal duplicates are skipped with an explicit count. Same identity with different meaningful content is a conflict, which aborts the whole batch. Any other processing error also rejects the full batch, so no apparently complete total is assembled from whichever files happened to work. Different report identities can have overlapping or touching periods; those receive a warning rather than being silently discarded. Their totals remain observations, not unique-email counts. The report comparison guide develops the distinction between duplicate reports and related traffic represented by separate reports.
Compare policy_published along with the period and reporter. A legacy pct value records historical policy information, while np, testing, and discovery_method can describe newer reports; RFC 9989 removed the old pct mechanism. Do not rewrite a report to match today’s policy or assume the analyzer fetched current DNS. If you compare Google feedback or Yahoo feedback with Microsoft observations, first establish whether the windows and sending streams are comparable. Different receivers need not observe the same traffic. A difference is a useful investigation question, not a provider ranking, proof of missing delivery, or reason to merge all observed instances into one deduplicated campaign total.
Keep the review temporary and the conclusion specific
A supported batch has no more than 10 files, with 4 MiB per source file and 16 MiB total source. Expanded XML is separately limited to 4 MiB per report and 16 MiB across the batch. Row, node, depth, and leaf-text limits apply alongside those byte limits, not instead of them. Read the complete contract on the DMARC Report Analyzer main page before assembling a large collection. A compressed file can fit its source allowance and still exceed an expanded or structural bound. Do not delete rows to make a complete-looking original fit. An intentionally smaller selection of whole reports must be described as that narrower collection.
The genuine analyzer uses an opaque-origin sandboxed frame, no-store responses, and a content security policy with connect-src none. A local terminable worker has a 12-second deadline. There is no upload, live DNS API, Microsoft account connection, or remote authentication test. Clear stops work and wipes the file input and selected files, pasted text, state, and every derived result, including hidden details and raw XML. Hidden-page changes and pagehide, pageshow, and freeze do the same, so switching away can end the session. There is no export, Copy, PDF, history, storage, or project feature. Keep any required original in an authorized location outside the tool rather than treating the display as a case record.
These controls do not promise secure memory erasure or protection from extensions, screen recording, compromised devices, clipboard copies, or a hostile replacement interface. Aggregate metadata can still disclose operational domains, infrastructure, contact information, and comments; there is no automatic redaction. Share only what an authorized recipient needs under your own rules. A useful closing observation names the selected report, its period, the relevant count, the supported alignment values, and what remains unknown. Reading the report does not change a tenant setting, send mail, connect an account, or start monitoring. Clear when finished, and use the Tools directory when the next task needs a different kind of evidence.
Questions about Microsoft DMARC Reports
Does Microsoft 365 send aggregate and forensic reports?
Microsoft documents aggregate reporting for valid rua=mailto destinations when the receiving Microsoft 365 domain’s MX points directly to the service, and says it does not send forensic reports. This does not promise complete coverage of every Microsoft-related message.
Can an Outlook message or message-trace result be analyzed here?
Supply the aggregate XML attachment, optionally in supported GZIP or single-XML ZIP packaging. An Outlook item, notification body, message headers, and trace table are not aggregate XML. Single-message headers belong in Email Header Analyzer, not this textarea.
Does a DMARC pass reproduce a Microsoft Defender decision?
No. This tool derives status from reported alignment fields and displays dispositions separately. It does not calculate composite authentication, apply tenant policies, inspect quarantine, or determine inbox placement. Raw authentication passes alone are not aligned DMARC passes.
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.
A report is not a tenant connection
Read the supplied XML locally
Return to the analyzer for supported files or a public example without opening Microsoft accounts or starting monitoring.