List Segmentation: Seven Segments Almost Everyone Needs

Segmentation that works is built on behaviour rather than demographics. The seven segments below — activity, purchases, tenure — assemble from data you already hold and require no surveys or extra form fields. Here is what each one is and what to send it.

Why demographics do not work

Start with what does not work. Splitting a list by gender, age and city looks like an obvious first step and nearly always disappoints.

The reason is that those attributes correlate weakly with behaviour. Two thirty-year-old women in the same city can behave in opposite ways: one buys monthly, the other subscribed for a promo code and never opened anything since. Sending them the same email is a mistake, and gender and age do not fix it.

There is a practical problem too: demographic data has to come from somewhere. That means a long signup form, and every extra field lowers signups. You pay in subscribers for data that then performs poorly.

Behavioural attributes, by contrast, accumulate on their own. Opened, clicked, bought, absent for three months — all recorded without asking anyone a question.

Seven behavioural segments

The seven segments below cover the needs of most email programmes. They are not mutually exclusive: someone can be both active and a regular buyer.

Note that the first four are built purely on activity and the last three on purchases. If you have no purchase data, the first four still work.

Activity outweighs every other attribute

If you implement only one attribute, make it activity. It affects both campaign results and the deliverability of the entire list.

The mechanism: mailbox providers judge a sender by recipient reaction. If half your list has ignored you for years, that drags inbox placement down for everyone else, including the people who read you regularly.

Splitting by activity solves both at once. Active contacts get the normal frequency, cooling ones get less and only the most valuable, inactive ones go into a re-engagement flow and are sunsetted afterwards.

The activity threshold depends on your sending frequency. At weekly sends, 90 days of silence is already a signal; at monthly sends it is normal. Count missed emails rather than calendar days: five or six consecutive sends without a single action is the boundary at any frequency.

Buyers and one-time buyers

The second most important attribute is whether someone has purchased. The difference between a subscriber and a buyer is fundamental, and the content for them should differ.

A subscriber who has not bought needs to be told why you matter. A buyer does not — they tried it. They need a reason for the next purchase, and that reason is a different thing entirely.

One-time buyers deserve their own segment — one purchase, made a while ago. It is the most underrated group: they already trusted you with money once, so the trust barrier has been cleared.

A practical device for them is offering complements to what they bought rather than new arrivals in general. Someone who bought a coffee machine cares more about beans than about a sale on cookware.

Regulars do not need discounts

Regular buyers, contrary to common practice, should not receive discounts more often than everyone else.

The reason is simple: they buy anyway. A discount for them is margin lost on a sale that would have happened regardless. Worse, regular discounts train people to wait for offers, and someone who used to buy at full price stops doing so.

What works instead: early access to new items, closed offers, unconditional free shipping, personal recommendations. All of these feel like a privilege and do not devalue your pricing.

Where to start with no segments at all

If there are no segments at all and the whole list receives identical emails, do not implement seven at once. Start with two.

• split the list into active and inactive using the past 180 days
• keep writing to the active ones as before
• send the inactive ones a three-email re-engagement flow
• exclude those who do not respond from regular campaigns
• check deliverability a month later: it will improve

That split takes an hour and delivers more than any subsequent refinement. Further segments get added one at a time, once the first is working.

When segmentation hurts

And an honest warning: segmentation can hurt. There are two situations where it should not be implemented.

• a list under a couple of thousand addresses: segments become too small to draw conclusions from
• no capacity for different content: five segments mean five different emails, not one with five audiences

The second is more common. Segmentation without differentiated content is pointless: if all five segments get an identical email, you have simply split one campaign into five and made more work for yourself.

And a note on measurement: you cannot judge the effect of segmentation by open rate, which is distorted by mail privacy protection. Look at clicks and revenue per email broken down by segment — the difference shows there.

In Mailvex an email for different segments is produced by copying: the base template duplicates, the product selection and copy change, and the brand kit stays shared. Synced modules let you change the footer across every variant simultaneously.

Frequently asked questions

At what list size does segmentation make sense?

From roughly two thousand addresses. On a smaller list the segments become too small: differences between them will be chance while the work multiplies. Below that threshold, splitting into active and inactive is enough.

Why segment by activity rather than just writing to everyone?

Because the inactive part of a list drags deliverability down for everyone else. Providers judge a sender by recipient reaction, and silent addresses worsen inbox placement even for people who read you regularly.

Should regular buyers get more discounts?

No. They buy anyway, so a discount is margin lost on a sale that would have happened regardless. What works is early access, closed offers and free shipping: those feel like a privilege and do not devalue your pricing.