Delivered is not the same as read

Why delivery rate flatters everyone

SMTP acceptance and inbox placement are decided at different moments by different systems. The receiving MTA accepts the message during the SMTP conversation, mostly on the basis of whether the address exists and the connection behaved. Filtering happens afterwards, inside the mailbox, where reputation, content and engagement history decide the folder. A message can be accepted with a clean 250 and be sitting in Junk two seconds later.

This is why senders describe the same experience over and over: the dashboard says 99% delivered, the reply rate is 0.2%, and nobody can explain the gap. There is no gap. The mail was delivered, to the spam folder.

How the test works, step by step

1. Warmerly selects seed targets: up to two on the same provider family as your mailbox, and up to two on different ones, so you get both the same-provider and cross-provider picture.
2. It sends one message per target from your mailbox's own SMTP connection, carrying a unique token in the subject.
3. It waits: long enough for the receiving side to finish filtering and file the message.
4. It connects to each seed mailbox over IMAP and searches Inbox, Promotions, Spam and Junk for the token.
5. It scores each result and averages them into a placement figure, recorded against the mailbox and trended over time.

A sweep runs automatically each day over a small batch of accounts that have not been tested in the last twenty-four hours, and you can trigger a test on demand from any mailbox page. Runs that hang are marked failed rather than left pending forever, so a stuck worker never masquerades as a result.

Same-provider and cross-provider, separately

Providers do not agree with each other. It is entirely normal for a domain to place at 96% into Gmail and 88% into Outlook, and the reasons differ: Microsoft weights IP reputation and its own SmartScreen history heavily, Gmail leans harder on domain reputation and engagement. A single blended placement number averages those two into a figure that describes neither.

Because Warmerly seeds across provider families, an Outlook-specific problem shows up as an Outlook-specific problem. That matters operationally: if 70% of your target list is on Microsoft 365 and your Outlook placement is the weak one, that is the number your pipeline actually depends on.

Degraded runs, and why we tell you about them

Real placement measurement needs real seed mailboxes at the major providers. When none are available, the only remaining targets are Warmerly's own warmup pool mailboxes, and those accept essentially everything, so the resulting score is close to 100 no matter how badly the sender is actually performing.

There is a strong commercial temptation to publish that number anyway. Warmerly does the opposite, in three enforced ways. A degraded run never writes to your health score. It is labelled in the interface as limited accuracy. And it is excluded from every downstream calculation that feeds a number, spam scoring, sending capacity, and the public placement counter alike. A trend comparison is only drawn between two runs of the same kind, because comparing a degraded run to a real one measures which targets were available, not a change in your placement.

Reading the heatmap

A single placement percentage hides the shape of the problem. The heatmap on the overview plots placement day by day, and the patterns it exposes are diagnostic. A uniform block that suddenly dims on one day usually means a volume spike or a content change on that day. A gradual fade across a fortnight is reputation drift. A single pale column against an otherwise solid grid is very often the day someone imported an unverified list.

The number tells you that something is wrong. The shape tells you when it started, which is most of the way to knowing what caused it.