Email segmentation is the practice of dividing a list into groups that receive different messages, based on something you know about them. The reason to do it is not sophistication. It is that a message written for everyone persuades no one, and that sending to people who are not interested is the fastest way to lose your inbox placement.
There are hundreds of ways to slice a list. Six of them are worth your time, and they are worth it roughly in this order.
1. Engagement — the one that protects everything else
Split your list into engaged (opened or clicked in the last 90 days), lapsing (90 to 365 days), and dormant (over a year). This is the first segment you should build and the only one that is not optional.
The reason is deliverability rather than relevance. Mailbox providers infer whether people want your email from whether they interact with it, and a list where 60% never open drags down inbox placement for the 40% who do. Suppressing the dormant group — not deleting, suppressing — typically lifts open rates on the remainder within two sends, and the lift is real rather than an artefact of a smaller denominator, because inbox placement genuinely improved.
Run the dormant group through a win-back campaign once, then stop mailing them. Details on the cleanup itself are in how to clean your email list.
2. Lifecycle stage — the one that changes what you say
Where someone sits in their relationship with you determines what a useful email looks like. A subscriber who joined yesterday needs orientation. A customer who bought last week needs onboarding and reassurance. A customer who bought fourteen months ago needs a reason to come back.
The minimum useful version is four stages: new subscriber, engaged non-customer, active customer, past customer. Each gets a different default message, and campaigns get written to a stage rather than to the list. Most of the revenue lift people attribute to "personalisation" is actually this. There is a full map in lifecycle email marketing.
3. Purchase behaviour
For anyone selling something, this is where the money is. The splits that matter are: number of purchases (one versus repeat), recency, category bought, and average order value. A one-time buyer from three months ago is the single most convertible segment most businesses have, and most businesses send them the same newsletter as everyone else.
The formal version of this is RFM segmentation — recency, frequency, monetary value — which sounds academic and is genuinely worth the hour it takes to set up.
4. Source
Where someone came from predicts what they want more accurately than almost anything else you know about them. Somebody who joined from a lead magnet about deliverability wants deliverability content. Somebody who joined at checkout wants product news. Somebody who came from a quiz told you their answers — use them.
This costs nothing to implement because the data arrives at signup. Store the source on the subscriber record from day one; you cannot backfill it later.
5. Stated preference
Ask people what they want and honour it. A preference centre that lets someone choose topics and frequency does two things at once: it produces a segment, and it gives people an alternative to unsubscribing. Offering "get less email" as an option on the unsubscribe page routinely retains a fifth of the people who would otherwise have left.
The caveat is that stated preference and revealed preference diverge. People say they want less promotional email and then click the promotional email. Treat stated preference as a constraint, not as a targeting signal.
6. Behavioural triggers
The last one is less a segment than a set of rules: viewed a page and did not convert, abandoned a cart, used a feature, hit a usage limit. These fire per person rather than per campaign, which is why they belong in automation rather than in your weekly send.
They convert far better than broadcasts because the timing is theirs rather than yours. They are also the most work to build, which is why they come last on this list rather than first. Behavioral segmentation for email has the full treatment.
How far to take it
There is a real cost to segmentation, and it is not technical. Every segment is a version of the email someone has to write, check and maintain. Four segments you actually customise beat twenty you populate with the same content.
A workable steady state for most small teams is: suppress the dormant, split by lifecycle stage into three or four groups, and run two or three behavioural triggers. That is achievable in an afternoon and captures most of the available lift. Everything past it has diminishing returns until your volume justifies the labour.
Measure it properly, too. Segmented campaigns look better on open rate almost by construction, because you removed the people who never open. The number that tells you whether segmentation worked is revenue per recipient, not open rate — see email conversion rate for how to calculate it without flattering yourself.
Frequently asked questions
What is the most effective email segmentation strategy?
Engagement-based suppression, because it improves results for every other segment simultaneously by raising inbox placement. Lifecycle stage is the most effective for revenue, but it only pays off if your email is actually reaching people, which is what engagement segmentation protects.
How small can a segment be before it is not worth it?
Small segments are fine for automated flows, which run continuously and accumulate volume over time. For one-off campaigns, the practical floor is whatever number makes the writing worth it — usually a few hundred people, unless their average order value is high.
Does segmentation actually increase revenue or just open rates?
Both, but not by the same amount and not for the same reason. Open rate rises partly because you excluded non-openers, which is arithmetic rather than improvement. Revenue rises because the message matches the moment. Track revenue per recipient across the whole list before and after, which is the only comparison that cannot be gamed by excluding people.
How many segments should I have?
Fewer than you think. Every segment is content someone has to write. Three to five well-maintained segments outperform fifteen that all receive the same email with a different first line.
