Sign in Get up to 500 free credits

Email feedback loops explained

A complaint feedback loop (FBL) is a service in which a mailbox provider tells a sender when its users mark the sender's mail as spam. Most FBLs send one report per complaint in the Abuse Reporting Format (ARF), a standard message format. Yahoo and Microsoft send such reports; Gmail does not, and instead shows aggregated complaint data in Postmaster Tools. Treat every report as an unsubscribe: suppress the address at once, then find the campaign or sign-up source that caused it.

How a feedback loop works

RFC 6449, "Complaint Feedback Loop Operational Recommendations", writes down practices that grew up in the industry over more than a decade. It began as a document of the Messaging Anti-Abuse Working Group (MAAWG) and is informational, not a standard. In its terms:

Section 1 explains how a provider knows whom to send a report to: usually by the IP address that delivered the mail or by the domain in its DKIM signature, because both are hard to forge. That is why Microsoft's program works by IP and Yahoo's by DKIM domain. The RFC also notes that complaints are forwarded without human review, so a few reports will not be actionable, and that providers may redact data such as the recipient's address.

The Abuse Reporting Format

RFC 5965 defines the report format, and RFC 6650 updates it with rules for how to use it. Section 2 of RFC 5965 describes a report as a multipart/report message with report-type=feedback-report and three parts:

  1. A human-readable explanation.
  2. A machine-readable part of type message/feedback-report.
  3. The original message (message/rfc822) or only its header block (text/rfc822-headers).

Each report covers one message. A shortened example:

Content-Type: multipart/report; report-type=feedback-report; boundary="b1"

--b1
Content-Type: text/plain

This is a spam complaint about a message from 192.0.2.1.
--b1
Content-Type: message/feedback-report

Feedback-Type: abuse
User-Agent: ExampleFBL/1.0
Version: 1
Original-Mail-From: <bounces+c42@mail.example.com>
Arrival-Date: Thu, 5 Nov 2026 14:02:11 +0000
Source-IP: 192.0.2.1
--b1
Content-Type: text/rfc822-headers

Message-ID: <c42.m981.r55017@mail.example.com>
From: Example News <news@example.com>
Subject: Your November offers
--b1--

Section 3 of RFC 5965 makes three fields required in the second part: Feedback-Type, User-Agent and Version. For complaint loops, RFC 6650, section 4.3, says reports should use Feedback-Type: abuse and should include Original-Mail-From, Arrival-Date, Source-IP and Original-Rcpt-To when the values are known. Section 4.5 adds two rules for you as the receiver: do not reject a report just because its feedback type is unfamiliar, and accept reports in which user data has been obscured, as described in RFC 6590.

In practice, the third part is where you find out what was reported. RFC 5965 says to rely on it first, since the fields in the second part are the provider's assertions.

Which providers offer one

ProviderProgramKeyed onWhat you get
Yahoo (and the domains it hosts, AOL included)Complaint Feedback LoopDKIM domain (d=)One ARF report per complaint
Outlook.comJunk Mail Reporting ProgramSending IPOne ARF report per complaint, headers only
GmailFeedback Loop in Postmaster ToolsFeedback-ID header and DKIM domainAggregated rates per identifier, no individual reports

Yahoo

Yahoo's Complaint Feedback Loop supports only DKIM-signed mail and is domain based. You create a Sender Hub profile, verify your domain and enroll it. Its FAQ says the DKIM domain in the d= tag must be enrolled, that reports go to any address you control, that IP-based reporting is no longer offered, and that an email service provider will generally enroll your domain for you. Yahoo's sender best practices ask for an active loop on every DKIM domain.

Microsoft

JMRP covers complaints from Outlook.com users and is tied to IP addresses you control through Smart Network Data Services. In 2026, reports changed to headers only, and practitioners report that the recipient's address is hidden. Setup, data and the changes are in Microsoft SNDS and JMRP.

Gmail

Google's Feedback Loop works differently. You add a header to each message:

Feedback-ID: campaign42:customer7:newsletter:examplesender

The last field is a mandatory sender ID of 5 to 15 characters; up to three optional identifiers come before it, such as campaign, customer or mail type. The message must be DKIM-signed, after the header is added, by a domain you have verified in Postmaster Tools. Google reports identifiers with an unusual spam rate in a Postmaster Tools dashboard, and only once an identifier has enough volume and enough distinct complaints. Google warns against per-message values such as a unique message ID, because they produce no useful totals. The data covers @gmail.com recipients and is totaled per identifier, not per person, so it shows which campaign has a problem but not whom to remove. See Google Postmaster Tools.

How to process reports

Prepare before you enroll

RFC 6449, section 4.1, lists what to have ready: access to your abuse@ and postmaster@ addresses, because providers send confirmation links there, and a dedicated address for reports. Section 4.4 suggests a separate alias per feedback loop, so you always know which provider a report came from.

Make every message traceable

Providers may remove the recipient's address, so put your own identifier in each message when you send it. RFC 6449, section 4.4, suggests the local part of the Message-ID, as in <c42.m981.r55017@mail.example.com> above: campaign 42, message 981, recipient 55017. A VERP return path carries the same information, but some reports do not include it. Avoid identifiers that look like an email address, since redaction may remove them.

Suppress the complainer at once

RFC 6449, section 4.3.1, tells senders of commercial mail to treat a complaint like an unsubscribe. Stop mailing that person from the list, and if you run several segmented lists, suppress them across all of them. The RFC calls suppressing the address across a whole ESP, for every customer, likely too broad. Put the address on your suppression list so that an import or a sync from your CRM cannot add it back.

Transactional mail is the exception that needs judgment. If someone reports a receipt or a ticket they bought, the RFC suggests deciding case by case, for example by asking them through your website whether they still want those messages. See transactional vs marketing email.

Look for the source

Section 4.3.2 explains how to read the trend. Count reports per day, campaign and list. A spike from new subscribers points at how addresses were collected; a spike from long-time subscribers points at one message, its content or its branding. Check whether the people who complain signed up through one form, one partner or one import.

Complaint rate thresholds

ProviderPublished limit
GmailKeep the rate below 0.10% and never reach 0.30% (email sender guidelines)
YahooKeep the rate below 0.3% (sender best practices)
Outlook.comNo published number; complaints are a main reputation factor

Your own figure will not match theirs exactly. Yahoo calculates the rate from mail delivered to the inbox, and Google counts messages that reached the inbox and were then marked as spam. RFC 6449, section 4.3.2, shows why the difference matters: 10 complaints on 10,000 sent messages look like 0.1%, but if only 500 reached the inbox, the provider sees 2%. It also notes that complaints are counted on the day people read the mail, so a quiet weekend after weekday sends can show a rate above 100%.

How the complainer reason relates

A feedback loop tells you after the fact that one of your recipients complained. Some people often report legitimate mail as spam. A check flags addresses believed to belong to such people with the reason complainer: the status is valid, because the mailbox accepts mail, and the action is remove. Removing them before you send avoids complaints you would otherwise only see in a report. The page on why complaints hurt explains what reports do to your reputation.

Neither tool replaces the basics. Many complaints come from people who do not expect your mail or cannot find the way out, so confirm sign-ups with double opt-in and offer one-click unsubscribe.

Questions

Do I need to enroll if my email service provider handles complaints?

Often not. Yahoo's FAQ says providers generally enroll their customers' domains and process the complaints. Ask yours whether it does, and whether it shows you the complaint counts.

Why do I get fewer reports than Postmaster Tools suggests?

Gmail sends no individual reports, and Microsoft and Yahoo send them only for the IPs or domains enrolled. Mail that was filtered to spam also draws few complaints, so a low count is not proof that recipients are happy.

Should I email a complainer to ask why?

No. RFC 6449 notes that senders rarely contact the person who complained, and that person has already told you they do not want your mail. Suppress the address and fix the cause.

Try it

Check a list with bulk email verification and remove the rows whose action is remove, or try one address in the free email checker.

Sources

How we wrote this: drafted by the email-validation.org team with the help of AI writing tools, from our product documentation and the sources linked in the text.

Get up to 500 free credits. No card needed.

New accounts get 500 free credits (100 when you sign up with a free email provider such as Gmail, Outlook or Yahoo, or another personal mailbox such as a paid one like Fastmail or one from your internet provider). Try it on your own list before you buy.

Create your account Open the free email checker

Updated · Maintained by the email-validation.org team