Waking up a dormant list without waking up your bounce rate
Somebody has looked at the list size, noticed a third of it hasn't opened anything in eighteen months, and proposed a win-back campaign. It's a reasonable idea attached to the riskiest send you can make.
The risk is specific. Dormant addresses are the population recycled spam traps are drawn from, and a large send to people who have shown no interest produces complaints at a higher rate than anything else you do.
The order the work happens in
Define dormant with a date, not a feeling
No open, click or purchase since a stated month. Writing the definition down is what makes the result measurable and stops the segment quietly expanding.
Verify the whole dormant segment
This is the one case where checking a segment you are unsure about is unambiguously worth it, because the failure rate in dormant data is high enough to change the decision.
Send to a sample of a few thousand first
Draw it randomly across the segment rather than from the newest end. The sample's bounce and complaint rates are your evidence for whether the rest goes out.
Read the complaint rate, not the opens
Above roughly 0.3% complaints, stop. That figure does more damage than the campaign can recover, and it is the number most people forget to look at on a win-back.
Sunset whatever does not respond
The purpose of a re-engagement campaign is to earn the right to stop mailing people. If it is not followed by removals, it was a normal campaign to a bad segment.
What each result means here
The same four results and their flags, read against this job. A catch-all worth keeping in one situation is one to exclude in another.
| Result | What to do with it |
|---|---|
| Valid | Include in the sample. A working mailbox is the precondition, not the goal. |
| Invalid | Remove permanently. On a dormant segment this group is usually larger than anyone expects. |
| Catch-all | Exclude from a win-back. You cannot confirm the mailbox and the person has not responded in a year, so there are two unknowns stacked. |
| Unknown | Exclude from this campaign. A dormant segment is the wrong place to spend reputation on a non-answer. |
| Role flag | Exclude. A role address that has been silent for a year is a shared mailbox nobody is reading. |
| Disposable flag | Remove. A throwaway address has certainly been abandoned. |
How you know it is finished
The sample came in under 2% bounces and under 0.3% complaints, the campaign has run, and everyone who did not respond has been moved out of the sending list rather than left for next year.
What this does not fix
Dormant segments carry more unknowns than fresh ones, and unknown results are not billed, so the bill on this job usually comes in under the segment size. A 50,000-address dormant file is $89 of credits at most.
Questions people ask
Is it safe to email a list I have not mailed in two years?
Not without verifying and sampling first. Two years of decay plus the chance of recycled spam traps makes it the highest-risk send most teams ever make, and the damage lands on the domain you use for everything else.
How large should the test sample be?
A few thousand, drawn randomly across the segment rather than from the newest records. A sample skewed to the freshest addresses measures the easy part and tells you nothing about the rest.
What complaint rate should stop the campaign?
Around 0.3%. Gmail and Yahoo both publish that figure as the line for bulk senders, and crossing it costs more than any win-back campaign returns.
Try it on the file in front of you
100 free checks a month, no card. Addresses we could not get an answer on come back labeled as unknown, and those are not billed.