How to Choose Your First Segments: Start With Five
How to choose your first segments: start with five, define each in one sentence, build the rule narrow, and add nothing until one is plainly too broad.
To choose your first segments, start with five: new and not yet contacted, engaged, cooling, dormant and customers, and add nothing else until one of them is plainly too broad to write a single message for. Define each in one sentence before you build the rule.
The instinct is either to build no segments at all or to build thirty. Both fail. This post walks through the five, how to define them, and the habits that keep a small scheme small.
How to choose your first segments: the five to start with
Most early businesses can run almost everything from a handful of groups. Each of the five has a distinct need.
| Segment | Who is in it | What they need |
|---|---|---|
| New, not yet contacted | Never sent anything | A welcome rather than an offer |
| Engaged | Opened or clicked recently | Your strongest material and your direct asks |
| Cooling | Last real engagement a while ago | Your most useful content, less often |
| Dormant | Received a lot, engaged with nothing for a long time | One honest win-back, then a rest |
| Customers | Already bought | Entirely different messages |
Add a fit based segment for your best target if you have enough contacts to make it meaningful. See how to segment by fit and behaviour for how fit is defined. Then stop.
Why five and not fifty
Segmentation multiplies work. Every segment is a message to write differently, a definition to maintain and a number to interpret. A company with a few hundred contacts and one offer needs three or four segments, not a dozen.
Over-segmentation is a real failure with two symptoms: groups too small to learn anything reliable from, and groups too numerous for anyone to remember. The test for adding a segment is that an existing one is plainly too broad to write a single message for. Until then, the sixth segment is administration.
Write the definition in one sentence first
Before you touch a rule builder, write the sentence. The guide's example reads leads tagged as interested in the advanced offer who visited the pricing page in the last two weeks. If you cannot say it in a sentence, you do not know what the segment is for, and the rule will be an accident of whichever filters were nearest.
Here is how the same discipline applies to the starter set, built only from the descriptions above.
- New: contacts we have never sent anything to.
- Engaged: contacts who opened or clicked recently.
- Cooling: contacts whose last real engagement was a while ago.
- Dormant: contacts who have received a lot and engaged with nothing for a long time.
- Customers: contacts who have already bought.
Words such as recently and a while ago must be turned into numbers of days by you. The guide gives no figure, so choose one that suits your sending rhythm, write it down and apply it the same way every time.
Name the segment for the rule
Name the segment for what it contains, not for the campaign it was made for. Advanced interest with a recent pricing visit will make sense in six months. October push will not, and someone will reuse it assuming it means something it does not.
This is the same reason a segment for a finished campaign should be deleted. Names tied to campaigns outlive their meaning and invite misuse.
Prefer all of to any of
All of narrows. Any of widens, often far more than expected, because one loose condition lets almost everyone in. When a segment comes back surprisingly large, a stray any of is the usual cause. Build narrow and widen on purpose.
The choice between a rule and a frozen list matters here too. Every one of the five is a dynamic segment, since you want whoever qualifies today. Frozen lists are for moments, as covered in dynamic vs static segments.
Give each segment a different message
Sending the identical newsletter to five segments is not segmentation. It is the same broadcast with extra steps. Each segment wants a distinct move.
| Segment | The distinct move |
|---|---|
| Engaged | Your strongest offer first, a request for a referral or testimonial, an invitation to try something new |
| Cooling | Your single most useful piece of content, a question about what they want to hear, and less frequent mail |
| Dormant | One clear win-back, then silence |
| New | A welcome that says what they will receive and how often |
| Opens but never clicks | One clear call to action near the top, not more content |
| Customers | Onboarding, expansion, requests for references, early access |
Each is a better message than the general one, and none needs a bigger list. If you cannot imagine writing a different message for a proposed segment, that is a reason not to create it.
Look at both numbers before the first send
The number who match a segment is never the number who should receive a message. Between the two sit people who never agreed to marketing, people who unsubscribed, addresses that bounced, contacts with no address, and duplicates sharing an address.
A count that ignores them is a promise you will not keep: the report says one number, the delivery another. Look at both before every send. If the gap is large, find out why first. A large gap is usually the base telling you it needs cleaning, and how to prevent duplicate contacts covers one of the biggest causes.
Keep test contacts out and consent in
Team addresses and test runs open and click everything, so they score as your most engaged contacts and distort the engaged segment. Mark them with one consistent fact and exclude them by rule, not by memory.
Consent is the opposite case. Whether someone agreed to receive marketing is the first filter on every marketing segment, so record it as a structured fact from the moment they arrive and make every marketing audience respect it automatically.
When to add a sixth
Add a segment only when one of the five is too broad for a single message. A useful way to notice is to try to write the message. If you find yourself writing two different openings, or choosing which half of the group to disappoint, that group needs splitting.
When you do add one, repeat the same steps: one sentence, a name for the rule, a decision about dynamic or static, and a distinct message. And review the whole set monthly, deleting any segment nobody can explain.
Checklist
- I have five segments or fewer, each with a one sentence definition.
- Each is named for its rule, not its campaign.
- Conditions are combined with all of unless I have a reason to widen.
- Each has a message written for it.
- I have looked at who matches and who will actually receive.
- Test contacts are excluded and consent is applied by rule.
A test for any segment you are about to add
Before you build a new one, answer these in writing. A segment that cannot answer them is a saved filter with a name.
- What is the one sentence definition?
- What will this group receive that the others will not?
- Will it change what someone receives, start something automatically, or make a number readable?
- Who is responsible for knowing it still means what it says?
- What would make me delete it?
The third question is the heart of it. A segment earns its keep when it does at least one of those three things. If it does none, leave it out.
What to do once the five are running
After the first send to each group, resist the urge to add more. Read the results first. Compare the result by segment, not just as one overall number, and note which segment responded best and which worst. The method is in how to measure results by segment.
The best performing group tells you who to find more of. The worst is either the cheapest thing to fix or the clearest thing to stop paying for. In both cases you have learned something from five groups that a dozen would have blurred.
Retire a segment as deliberately as you create it
A segment for a finished campaign, a rule on a tag that no longer exists, a test nobody looked at: each clutters the list and invites misuse. When nobody can say what a segment is for, delete it.
The five starting groups are not at risk, because each describes a standing state. The risk is the extra ones added later. Keep the total small enough that everyone knows what each one is for, and a short list that everyone understands will beat a long one nobody trusts.
Frequently asked questions
What if I only have a few hundred contacts?
Then use three or four of the five. A company with a few hundred contacts and one offer does not need more.
Should customers be a segment from the start?
Yes. They are your most valuable segment and the one most often sent the wrong things. Recording them as a distinct group is what lets you give them customer messages.
How do I define recently?
You choose the number of days, write it down and use it consistently. The guide offers thirty days as an illustration of a rule, not as a requirement.
What if a segment comes back far larger than I expected?
Look for a stray any of in the conditions. Switch to all of, check the count again, and widen deliberately. If the count still surprises you, read segmentation mistakes founders make for the usual causes.
The full method, with the checklists, is in The Segmentation Playbook, a free 35-page playbook.