Segmentation gets presented as an advanced discipline requiring a data team. In practice, most small businesses would see the majority of the available gain from one split: people who have bought from you, and people who have not.
Beyond that, three or four segments cover almost everything worth doing. The elaborate twenty-segment matrices in marketing guides mostly produce maintenance work and empty segments.
Quick Info
First segment to build
Customers vs non-customers
Second
Engaged vs unengaged
Third
By interest or service
Realistic number of segments
Three to five
Typical uplift from basic segmentation
Meaningfully higher clicks, lower unsubscribes
Sign of over-segmentation
Segments too small to send to
Why it works
Relevance is the whole mechanism. A more relevant email is opened more, clicked more, complained about less, and that improved engagement then improves delivery for everything else you send.
The compounding effect is the part people miss. Segmentation does not just improve the segmented campaign; it protects your sender reputation, which protects every future send.
Sending everybody everything is how you teach mailbox providers that your mail is not wanted.
Segment one: customers and non-customers
The highest-value split and the one most small businesses do not have. These two groups need entirely different email.
Non-customers
—Educational content, proof, case studies.
—Clear explanations of what you do and what it costs.
—Offers designed to produce a first enquiry or purchase.
Customers
—Complementary services or products, not the thing they already bought.
—Getting more out of what they have.
—Review and referral requests.
—Renewal or repurchase reminders at the right interval.
Sending an acquisition offer to somebody who bought last week is the most common and most irritating segmentation failure. It is also a one-condition fix.
Segment two: engaged and unengaged
Based on behaviour rather than attributes. This one protects your deliverability as much as it improves your numbers.
01Highly engaged: clicked in the last thirty days. Mail these more often, test new things here, and start any new sending domain with them.
02Engaged: opened or clicked in the last ninety days. Your normal audience.
03Lapsing: nothing in ninety to one hundred and eighty days. One re-engagement email.
04Unengaged: nothing in six months. Suppress after the re-engagement attempt.
Because open tracking is unreliable now, weight clicks more heavily than opens when building these. An open-based segment is partly fiction.
Segment three: interest or service
Which of your things they care about. Only worth building if you genuinely sell distinct things to distinct audiences.
01Tag at signup, by which form or lead magnet they came through.
02Tag by clicks: somebody who clicks web design links three times is interested in web design.
03Tag by page visits, if your platform connects to your site.
04Ask directly. A one-question email saying "which of these do you want to hear about?" works better than anybody expects.
05Then actually use the tags. Untouched tags are the most common form of segmentation theatre.
Segments worth adding if relevant
—Geography, for local businesses with multiple areas or events.
—Purchase value, so your best customers get different treatment from one-off buyers.
—Subscription date, so people who joined in the last month can receive the welcome content rather than a mid-conversation broadcast.
—Lifecycle stage for a longer sales cycle: enquired, quoted, booked, completed.
—Not: gender, age or job title, unless your offer genuinely differs by them. These usually add nothing.
How to actually build them
01Start with one condition. Purchased yes or no. Do not attempt a matrix.
02Use your platform's dynamic segments rather than static lists, so membership updates automatically.
03Confirm the data exists before building the segment. Most empty segments are missing data, not missing customers.
04Name them so a future you understands them: "Customers - purchased 12mo" not "Segment 4".
05Check the size. A segment of forty people is a personal email, not a campaign.
06Review quarterly and delete the ones you never send to.
What to actually send each
Segmentation without differentiated content is just extra admin. The point is to change the email.
—Change the examples. Same argument, examples from their sector.
—Change the offer. Acquisition offer for prospects, expansion offer for customers.
—Change the frequency. Mail your highly engaged segment more often and your lapsing segment rarely.
—Change the subject line, which is often the only change needed to lift clicks.
—Do not rewrite everything. One email, three variations, is a realistic workflow.
When segmentation backfires
—Segments too small to be worth sending to, which multiply your work and produce nothing.
—Segments based on data you do not maintain, which decay into inaccuracy.
—So many segments that sending becomes an hour's work and you send less often as a result.
—Over-personalisation that reads as surveillance rather than service.
—Segmenting instead of writing better email, which is the most common displacement activity in the discipline.
Three segments you use every send beat twenty you built once and never mailed.
What to do this week
01Build the customers versus non-customers segment.
02Build the engaged and unengaged segments on clicks, not opens.
03Suppress anybody with no clicks or opens in six months, after one re-engagement email.
04Send your next campaign to prospects and customers with one changed paragraph and one changed call to action.
05Compare click rates against your last unsegmented send.
That is an hour of setup and it is where most of the available improvement is. The deliverability side benefits immediately too.
Frequently Asked Questions
What is the most important email segment?
Customers versus non-customers. It is one condition and it delivers most of the available gain, because the two groups need completely different email.
How many segments should I have?
Three to five for most small businesses. Beyond that, maintenance grows faster than results and segments get too small to send to.
Should I segment on opens or clicks?
Clicks. Privacy features make open data unreliable in both directions, so an open-based segment is only partly real.
Does segmentation improve deliverability?
Yes, indirectly and substantially. More relevant mail gets more engagement and fewer complaints, which improves how providers treat everything you send.
Do I need a big list to segment?
No. Even a few hundred subscribers benefit from splitting customers from prospects. The tactics that need scale are predictive ones, not basic segments.
Is it worth asking subscribers what they want?
Yes. A single-question email asking which topic they care about produces better tags than most behavioural inference, and it gets replies.
Before You Go
Split customers from prospects, split engaged from unengaged on clicks, suppress the long-dormant, and change one paragraph and one call to action per segment.
That is an hour of work and it will do more for your click rate than any subject line experiment. The automations are where segments really earn their keep.
Two segments, used every send. Start there.
WRITTEN BY TARIQ SALLAM
Marketing Consultant. Entrepreneur. Content Creator.
I'm a marketing consultant, entrepreneur and content creator. I help businesses grow through practical marketing, websites, SEO, content and AI.
More About Tariq →
Keep reading