Engagement-Based Segments: Active, Lapsed, and At-Risk
The split that protects deliverability, with concrete thresholds and what to send each tier.
On this page
The short version
- The one segment every list should have. It costs nothing to define, needs no new data, and protects delivery for everyone who does engage.
- Define the tiers in sends rather than months. Twelve consecutive sends ignored means the same thing whether that took six weeks or two years.
- Clicks are the reliable signal. Opens are contaminated by privacy pre-fetching and should be supporting evidence rather than the definition.
Engagement segmentation groups subscribers by whether they are still paying attention. It is the highest-return split available and the one most often deferred, because its immediate effect is to reveal that a large part of the list is inactive.
That revelation is the point. Everything else in a programme is measured against a denominator that includes people who will never open anything, and correcting that changes what every other number means.
Define it in sends, not months
"No engagement in six months" means something entirely different for a daily sender than for a quarterly one. The first has ignored a hundred and eighty emails; the second has ignored two.
Counting sends removes that. Someone who has received twelve consecutive messages without clicking has told you something, and the calendar time it took is irrelevant.
Convert to dates for implementation, because that is what platforms segment on — but derive the date from a send count rather than adopting a number from an article.
Four tiers, and what each should receive
| Tier | Definition | Send them | Include in |
|---|---|---|---|
| Active | Clicked within the last 3 sends | Everything | Everything |
| Cooling | No click in 4-8 sends | Everything, best subject lines | Most campaigns |
| Lapsed | No click in 9-15 sends | Reduced frequency only | Not promotional sends |
| Dormant | No click in 16+, or never | A win-back, then removal | Nothing else |
The dormant tier should be excluded from promotional sends specifically. They are the least likely to buy and the most likely to report it as spam, which is the worst combination available.
What it does for deliverability
Mailbox providers score you on how recipients behave, and those are ratios. A third of the list never opening depresses every ratio, which affects placement for the two-thirds who do.
Suppressing the dormant tier raises those ratios immediately, because the denominator shrinks. Placement then improves over the following weeks for everyone else, which is the second and larger effect.
That is why this segment pays for itself before it delivers any relevance benefit at all. It is a deliverability measure that happens to be a segment.
Handling the never-engaged
Subscribers who have never clicked since signing up are a different case from those who used to and stopped. They may have mistyped an address, subscribed insincerely, or never received anything at all.
That last possibility is worth ruling out before removing them. If a cohort from one acquisition source has near-zero engagement while others are normal, the problem is likelier to be a broken delivery path than a group of uninterested people.
Check that before concluding. Then run a win-back, and remove what does not respond — the never-engaged are also the group most likely to contain recycled spam traps.
Keeping the tiers current
The tiers must recalculate continuously. A snapshot taken in January and used in June is describing a list that no longer exists, and it will suppress people who have since re-engaged.
Most platforms support rules that evaluate at send time, which is what you want. Where they do not, a scheduled recalculation weekly is adequate; monthly is not, because a subscriber can move two tiers in a month on a weekly send schedule.
Engagement segmentation check
- Tiers defined in consecutive sends, converted to dates
- Based on clicks, with opens as supporting evidence only
- Dormant subscribers excluded from promotional sends
- Never-engaged cohorts checked for a delivery fault first
- Rules recalculate at send time or at least weekly
- A win-back runs before anyone is removed
- Engaged subscriber count is the headline figure, not total
Frequently asked questions
Where should the thresholds sit?
The numbers above are a reasonable starting point for a weekly or fortnightly sender. Derive yours from where engagement actually falls off in your own data — plot clicks against consecutive sends and the shape will suggest the bands.
Should lapsed subscribers get fewer emails or none?
Fewer first. Reduced frequency recovers a meaningful share, and it is a smaller step than removal. Reserve suppression for the dormant tier.
Does this hurt total revenue?
Sending less to people who never engage costs almost nothing, because they were not converting. The placement improvement for engaged subscribers typically more than covers it, though it takes weeks to appear.