RFC 9991 keeps DMARC failure reports alive, but it does not turn them into a dependable replacement for aggregate reporting. The standard updates the failure-report format and gives domain owners more useful diagnostic fields. It also makes the privacy, abuse, and operational costs impossible to ignore.
For most domains, the sensible answer is not to enable ruf by default. Use aggregate rua reports as the operating system for DMARC, and enable failure reports only when there is a defined diagnostic or incident-response reason, a controlled destination, and a process for handling sensitive message data.
RFC 9991, DMARC Failure Reporting, updates RFC 6591 and works alongside the newer DMARC specification in RFC 9989. It describes reports about an individual failed message or a group of similar failures, rather than the periodic population-level view in DMARC aggregate reports.
The updated format can carry more explicit authentication context in the ARF report, including:
Identity-Alignment, identifying whether DKIM, SPF, or both failed to produce an aligned identityAuth-Failure: dmarc when the report describes a DMARC alignment failureThat is valuable when an aggregate row says only that a source failed. A failure report may help distinguish a broken legitimate sender from a message abusing the domain, without waiting for the next aggregate reporting interval.
The important qualification is that ruf remains a request. The receiving system decides whether it generates a report, what it includes, how it redacts it, and how frequently it sends it. A valid DNS record cannot make every mailbox provider send message-level reports.
An aggregate report normally gives statistical and authentication metadata. A failure report can contain selected headers or, in some cases, the complete failed message. That difference changes the risk profile.
Depending on the receiver and its privacy policy, a report may reveal:
This is not merely a theoretical concern. RFC 9991's privacy section specifically discusses PII, non-public information, mailing-list recipients, forwarding destinations, and the possibility of exposing complete messages. A legitimate message can fail authentication because of a forwarding or signing problem and then become the source of a sensitive report.
Failure reports can contain more than evidence about authentication. Treat them as potentially sensitive copies of messages, not as harmless log lines.
The report destination matters just as much as the published tag. A shared mailbox with broad access, automatic forwarding, weak retention controls, or no malware handling is a poor destination for ruf data. A third-party destination also needs a privacy and contractual review, not only a DNS check.
Aggregate reports answer the questions needed for normal DMARC operations:
They do this without normally carrying the contents of individual messages. They can be delayed and imperfect, but their lower sensitivity and better scalability make them the durable foundation for sender inventory, rollout, and policy decisions.
Failure reports answer a narrower question: what happened in this particular failure, and can the extra detail shorten an investigation? That makes them useful as an incident-detail channel, not as the primary measurement system.
The absence of failure reports is also weak evidence. Many receivers limit or disable them, and others may redact enough data that the report cannot reconstruct the original message. No ruf mail does not prove that all messages passed DMARC.
ruf is justifiedPublishing a failure-report destination can make sense when all of these conditions are true:
rua process, not compensate for the lack of one.For a short investigation, a temporary request is usually safer than leaving ruf enabled indefinitely. Record the start and end date, the hypothesis being tested, and the success criteria. Remove the request when the diagnostic value has ended.
ruf and fo fit togetherThe ruf tag identifies where failure reports may be sent. The fo tag describes which failure conditions may trigger them. For example:
v=DMARC1; p=reject; rua=mailto:aggregate@example.com; ruf=mailto:failure@example.com; fo=1Here, fo=1 requests a report when at least one underlying authentication mechanism fails to produce an aligned pass. Other fo values can request reports for both mechanisms failing or for a DKIM or SPF failure, depending on the value and the receiver's implementation.
Two details are easy to miss:
fo has no useful effect without a ruf destination.Do not interpret the record as a guarantee that every failed message will arrive at failure@example.com. It is a preference expressed by the domain owner within a protocol that gives the receiver discretion for privacy and operational reasons.
DMARCPal does not process ruf reports. Keeping aggregate rua reporting as the supported workflow preserves privacy and keeps normal DMARC monitoring focused on scalable evidence. Since failure reports are rarely comprehensive, most domains lose little operational visibility by leaving ruf out of their routine setup.
If the ruf URI is outside the domain's organizational domain, the receiving system must verify that the external destination has authorized the reports. RFC 9991 points to the external-destination procedure in RFC 9990, substituting ruf for rua.
For a domain such as example.com sending reports to reports.example.net, the destination needs the corresponding authorization record at the location defined by the DMARC reporting procedure. This protects the destination from being forced to accept a stream of reports by a domain owner it does not know.
The authorization check is necessary, but it is not the complete review. Also verify that:
When there is a real reason to test failure reporting, use a bounded rollout:
rua reports.fo option. Do not request more classes of failures than the investigation requires.ruf when the investigation is complete and keep rua as the long-term reporting channel.Receivers also have responsibilities. RFC 9991 calls out rate limiting and incident aggregation because an attacker can send large numbers of messages that impersonate a domain and trigger reports. A report generator should not turn the domain owner's request into an easy denial-of-service path.
| Situation | Recommendation |
|---|---|
| First DMARC deployment | Use rua; do not add ruf by default |
| Sender inventory and enforcement rollout | Use aggregate reports and controlled testing |
| One intermittent DKIM or forwarding failure | Consider a short, targeted ruf request |
| Active spoofing investigation | Consider ruf with strict volume and privacy controls |
| High-volume domain with no report-handling process | Do not enable it |
| Destination is a broadly accessible mailbox | Do not enable it until access is redesigned |
| No receivers are known to support failure reports | Do not depend on it |
RFC 9991 makes failure reports more useful to the people who can receive them. It does not make them universal, complete, or safe by default. The new diagnostic fields are a reason to use them deliberately, not a reason to publish ruf on every domain.
For most organizations, keep the operating model simple: collect and analyze rua, investigate individual failures through controlled test messages and message headers, and add ruf only for a documented case where its extra detail is worth the privacy and handling cost.
DMARC failure reports are a microscope. They can reveal the detail needed to explain one authentication failure, but they are not a census of all mail and not a substitute for aggregate reporting. Use them briefly, protect what they reveal, and remove them when the question has been answered.