Sunsetting Email Subscribers: The Right Way to Let Go
A sunset policy with a defined last attempt, and the arithmetic that makes removal profitable.
On this page
The short version
- A sunset policy is a written rule with a defined last attempt and a defined end, decided once rather than argued each quarter.
- The arithmetic usually favours removal: a smaller engaged list gets better placement, and better placement produces more clicks than the removed segment was contributing.
- The addresses most likely to have become recycled spam traps are exactly the ones a sunset policy removes, which is the argument that outweighs the list-size objection.
Sunsetting is the deliberate removal of subscribers who have stopped engaging, after a final attempt to bring them back. It differs from cleaning in that it is a standing policy rather than an occasional pass.
Having it as a policy matters. A rule agreed in advance gets applied; a decision taken each quarter gets deferred each quarter, because the immediate cost is visible and the benefit is not.
The arithmetic
Removing the inactive share improves two things at once. Engagement ratios rise immediately, because the denominator shrinks. And placement improves over the following weeks, because those ratios are what mailbox providers score you on.
The clicks lost from the removed segment are close to zero, since they were not clicking. The clicks gained come from better placement on the people who were.
Whether that trade is positive depends on your numbers, and it usually is by a comfortable margin. Work it through on your own figures before the conversation, because the person who owns the subscriber count will reasonably want the arithmetic rather than the principle.
The trade, worked
A 100,000-address list where 35,000 have not clicked in a year. Illustrative rates chosen to show the mechanism.
Illustrative model, not measured data. Run the same six rows on your own placement and click rates before removing anything.
Writing the policy
Four decisions, made once. The inactivity threshold, expressed in consecutive sends rather than months. The final attempt — how many messages, over what period. The end state: removed, or moved to a suppressed segment that receives nothing. And the review date for the policy itself.
Suppression is usually the right end state rather than deletion. It keeps the consent record and the history, stops the mail, and means a future import cannot quietly re-add someone who left.
Write the threshold down with its reasoning. "Twelve consecutive sends without a click" invites the question of why twelve, and answering it once in the document prevents the conversation recurring.
The last attempt has to end
Three messages, then stop, and stopping must be real. A sequence that announces a final message and continues has taught the list that your statements are approximate.
The one that recovers the most people is the middle message offering reduced frequency rather than an offer or a discount — most disengagement is volume rather than relevance, and that message addresses it directly.
After the sequence, apply the end state on a defined date. An indefinite grace period is the mechanism by which a sunset policy quietly becomes a document nobody follows.
Sunset policy check
- The threshold is written down, in consecutive sends, with its reasoning
- The final sequence is three messages and it ends
- The middle message offers reduced frequency
- The end state is suppression rather than deletion
- The removal date is defined and applied
- A backup exists before each run
- The engaged subscriber count is the headline figure
- The policy itself has an annual review date
Frequently asked questions
Where should the inactivity threshold sit?
Ten to twelve consecutive sends without a click is a reasonable default. Derive yours by plotting clicks against consecutive sends — where the curve flattens is where subscribers have effectively stopped.
What about subscribers who only read and never click?
How do I justify this to whoever owns the subscriber count?
With the arithmetic on your own numbers, and by changing the reported figure to engaged subscribers. A count including people who will never open anything is measuring storage rather than audience.