Three operators, three ways of telling you

by Francis Rozange | Sep 8, 2026 | Email Marketing

Abstract 3D render: three return channels of light, one wide, one narrow, one sealed

A recipient clicks “report spam”. That gesture decides your fate, and it happens without you knowing anything about it.

Unless you are enrolled in a feedback loop. All three major operators offer one. They do not give you the same thing, and the gap between them is considerable.

Yahoo gives you the address

Yahoo’s loop is the easiest to understand, because it does exactly what the name promises.

“When recipients mark your email as Spam, it negatively impacts your reputation as a sender. […] Yahoo sends the enrolled address a report in a format called ARF (Abuse Reporting Format)”

The report contains the reported message. So you know who clicked, and you can act: “suppress that recipient from further campaigns”.

The entry condition is technical and singular: “This service supports DKIM-signed email and is a domain based service. Senders must sign their outbound email with DKIM, so that Yahoo can determine the actual sender of an email.”

A DKIM signature, a sender profile, a verified domain, an enrolment. Nothing else.

Google never gives you the address

Google’s loop carries the same name and does not render the same service. It is reserved for large volumes, and it returns no addresses at all.

“If you’re a large volume sender, you can use the Feedback Loop (FBL) to identify campaigns in your traffic that are getting a high volume of complaints from Gmail users.”

The word is “campaigns”. You do not learn who reported you, you learn which campaign is being reported.

The mechanism rests on a header you must insert yourself, the Feedback-ID header, made of four fields:

Feedback-ID: CampaignIDX:CustomerID2:MailTypeID3:SenderId

Three free identifiers, plus a final mandatory one: “Mandatory unique Identifier (5-15 characters) chosen by the sender. It should be consistent across the mail stream.”

Google then aggregates on each of them independently, reading the first four fields from the right. Which leaves you to design your own analysis grid: by campaign, by customer, by message type.

One design mistake is enough to make the whole thing useless, and Google takes care to name it: “When choosing the Identifiers, the sender should not use a parameter that will be unique across every single mail.” One identifier per message reports nothing, ever.

The threshold Google does not publish

Here is the sentence that explains why so many enrolled senders receive no data at all:

“For a given day’s traffic, FBL reports are generated only if a given Identifier is present in a certain volume of mails as well as in distinct user spam reports.”

A certain volume, and a certain number of distinct reports. Neither figure is published.

That is a constant with these operators: they publish the rule and keep the threshold. Yahoo does the same with its own bulk sender threshold, and for the same reason: a published threshold is a threshold people deliberately skirt.

The technical prerequisites are listed without ambiguity. DKIM signature applied after the header is added, by a domain you control. That domain verified in Postmaster Tools. Sending IPs published in the signing domain’s SPF. Valid PTR records. Only one verified header per message.

Add the scope limit, written in one line: “FBL data will only pertain to @gmail.com recipients.” Your Google Workspace recipients are not counted.

Microsoft, or the documentation that vanished

Microsoft has long offered senders two services: the Junk Email Reporting Program, which relays complaints, and Smart Network Data Services, which exposes statistics by IP range.

Both still exist. Their public documentation no longer does, in readable form.

As of 8 September 2026, the historic addresses of the Outlook postmaster site redirect to a single page whose only textual content is the announcement of the authentication requirements of 5 May 2025. The pages covering sender services no longer render usable content.

We will therefore not describe what these programmes provide today: the primary source is not accessible, and we prefer to write that rather than copy a second-hand description.

The fact is worth noting in itself. The operator that hardened its requirements on volume senders is also the one whose sender tooling is the least documented.

What these three contracts imply

Put them side by side and one conclusion follows.

At Yahoo, you can handle a complaint individually: the address leaves the database, the incident is closed.

At Google, you can only handle a segment. A campaign above the threshold tells you that some content, some address source or some pace is a problem. You then have to work out which addresses, without ever being told.

Which puts the complaint rate back in its proper place. It remains the only number that decides, and a feedback loop is not a way to bring it down after the fact. It is a way to know which part of your programme is pushing it up.

What to do tomorrow morning

Enrol with Yahoo if you have not. It is free, the only condition is a DKIM signature you already have, and it gives you addresses to remove within the first week.

Add the Feedback-ID header to your Gmail traffic, designing it before implementing it. Choose your three identifiers according to the question you want to be able to ask in six months: by campaign if you test content, by address source if you doubt one of them, by message type if you mix streams.

Then check that DKIM signing happens after the header is added. That is the most common integration mistake, and it makes the whole thing silent without ever producing an error.

And expect nothing from the Microsoft side until its documentation returns. With that operator, your rejection logs remain the best source of information you have.

Sources

LaFactory works email on the evidence: headers, DNS records, rejection logs. No open rate promises, ever. Get in touch for a deliverability audit.