What Mailbox Providers Actually Check Before They Deliver Your Email

Mailbox providers decide where your email lands by checking three things in order: whether the message is authenticated, what reputation the sending domain and IP have built, and how recipients have behaved toward your mail before. Copy is evaluated last and matters least, which is why a well written email from a misconfigured domain still lands in spam.

Most deliverability advice starts at the wrong end. It begins with subject lines and spam trigger words, which sit at the bottom of the decision stack, and never addresses the two stages that reject mail before anyone reads a word of it.

Stage One: Is the Message Authenticated?

This is a pass or fail gate, and it happens before content is considered at all. Three records do the work.

SPF declares which servers are allowed to send mail for your domain. DKIM attaches a cryptographic signature proving the message was sent by someone holding your domain’s key and was not altered in transit. DMARC tells the receiving provider what to do when SPF or DKIM fails, and ties the result back to the domain in the visible From address.

The common failure here is not a missing record. It is misalignment: records that exist but do not match the domain actually sending, usually after a tool was added or a sending platform was changed and nobody revisited DNS. The mail is technically authenticated for the wrong identity, which from the provider’s view is indistinguishable from someone sending on your behalf without permission.

Nothing downstream rescues a message that fails this stage. Perfect copy to a perfectly targeted prospect from an unauthenticated domain is filtered before it is read.

Stage Two: What Reputation Does This Sender Have?

Reputation is the provider’s accumulated record of how your domain and sending IP have behaved. It is built over time and it is not transferable.

A brand new domain has no record, and the absence of a record is not treated as neutral. From the provider’s perspective, a domain that has never sent anything and then sends hundreds of cold messages in an afternoon matches the behavior of a disposable spam domain, because usually it is one. This is the entire reason warmup exists, and why a program that is fully built in 72 hours still produces meaningful volume only after the ramp. The mechanics are covered in What Is a Warmed Sending Domain?

Reputation also degrades, and it degrades faster than it builds. Sending to invalid addresses, triggering spam complaints, or pushing volume up in a step change rather than a ramp all move the number down. The asymmetry is the important part: damage accumulates in days and recovery takes weeks, which is why disciplined programs throttle deliberately instead of sending everything they are technically able to send.

Stage Three: How Do Recipients Behave Toward Your Mail?

This is the stage that decides placement for senders who clear the first two, and it is behavioral rather than textual.

Opens, replies, forwards and messages pulled out of the spam folder are positive signals. Deletions without opening, and especially spam complaints, are negative ones. Providers watch these per sender and adjust accordingly, which means your placement is partly determined by how well targeted your previous sends were.

That has a consequence most teams miss. Sending to a poorly matched list does not merely waste those sends. It teaches the provider that your mail is unwanted, which lowers placement for the well matched contacts you send to next month. A bad list is not a neutral experiment. It is a withdrawal from an account you need later.

Where Does Content Actually Fit?

Last, and narrower than its reputation suggests. Content filtering catches recognizable patterns: heavy link stuffing, attachments that do not belong in a first touch, formatting that mimics known bulk mail, and large identical sends with no variation.

Writing in plain language and keeping a first touch simple clears this stage comfortably. What content filtering cannot do is compensate for the two stages above it. This is the whole reason the diagnosis goes wrong so often: the only lever most teams know about is the one with the least influence, so they pull it repeatedly and conclude the channel is broken when the reply rate does not move. The same misdiagnosis in a different form is covered in Why Your Outbound Stalls in Month Three.

How Do I Diagnose Placement Problems in the Right Order?

Work down the stack, not up it.

Start by verifying SPF, DKIM and DMARC actually align for the domain in the From address, not for a domain that used to be in it. Then look at whether volume ramped or jumped, and whether the domain was warmed before campaign sending started. Then look at list quality, specifically what share of recent sends went to addresses that were never going to be a fit. Only after those three come back clean is it worth rewriting the email.

Most of the time the answer is found in the first two steps, which is also where it is cheapest to fix.

What Does This Mean for Choosing a Vendor?

It means infrastructure questions are the real questions. Ask who registers the sending domains and in whose name, since reputation you paid weeks to build should not stay behind when you leave. Ask how warmup is handled and how long it runs before campaign volume begins. Ask how the list is validated before the first send. A vendor who answers those three clearly is running a program. A vendor who only wants to talk about copy is selling the least important layer.

Autopilot registers domains in the client’s name, warms them before volume begins, and validates lists before sending, because those are the stages that decide placement. The ownership question specifically is covered in Who Actually Owns Your Sending Domains?, and the managed, human written version of the same infrastructure is Managed Email.