Segmentation on a small list is not about scoring or behaviour. It is three splits: what work someone had done, when they had it, and whether they are a customer at all. Those three produce almost all the relevance available and require nothing beyond your own job records.
Why the usual advice does not apply
Most segmentation guidance assumes a list large enough that dividing it still leaves useful groups, and behavioural data from frequent purchases.
A local service business has neither. Three hundred customers who buy every few years produces segments of forty, and there is no browsing behaviour to score. Applying the standard approach produces complexity with no benefit.
What does apply is that a message tied to something specific about the recipient outperforms a general one, and on a small list that specificity comes from your job records rather than from tracking.
The three splits that matter
What they had done
The most useful, because it determines what is relevant next. Someone who had a roof replaced does not need a roof message; they need gutters, or an inspection in five years.
This split is what makes a message read as attentive rather than as a mailshot, and it comes directly from invoices.
When they had it
Turns a list into a schedule. Anything with a service interval, a warranty period, or a typical lifespan produces a date on which a message becomes relevant to that one person.
This is the split that makes triggered messages possible, and triggered messages are what let you communicate often in total while each person hears from you rarely.
Customer or not
Someone who received a quote and did not proceed is a different audience from someone who has paid you. Different tone, different content, and a different basis for contacting them at all.
Keeping these apart also matters for consent, since the basis for contacting a past customer differs from the basis for contacting an enquiry that never became one.
What that produces in practice
| Segment | What they get |
|---|---|
| Customers, work due | A reminder tied to their own date. The highest-performing message you can send. |
| Customers, recent | Follow-up, review request, and nothing else for a while |
| Customers, dormant | Seasonal prompts and occasional useful information |
| Quoted, did not proceed | One seasonal message, at most, where consent supports it |
| Everyone | Two seasonal messages a year, tied to your demand pattern |
That is five groups from three fields, and it covers a service business comfortably.
Optional splits worth having
- Area, where you serve several and something applies to only one, such as a job you are already doing in that neighbourhood.
- Residential or commercial, which are different buyers wanting different messages.
- Job size, where a large project customer is worth a different conversation from a callout.
Add these only when you have a message that genuinely needs them. A segment created without a message to send it is administration rather than marketing.
What not to bother with
Engagement scoring, purchase frequency modelling, and anything requiring behavioural data you do not have.
Also worth skipping on a small list: splitting for the sake of split testing, since the groups will be too small for the results to mean anything, and personalisation beyond a first name, which does not improve on a message that is already specific about the work.
The one thing to track properly
Whether the recipient is still a customer, and whether they have opened anything in a year.
That second one is worth acting on rather than ignoring. Continuing to send to people who never open costs you in deliverability terms, and moving them to a group you contact once a year is better than either sending indefinitely or removing them.
It also keeps the reported figures honest, since a list padded with people who stopped reading two years ago makes every open rate look worse than it is, which is the wider point in how often to email without losing people.
Where the data comes from
Your invoicing system, exported with the address, the date, and what work was done.
That is the whole requirement. No tracking, no scoring, no additional software beyond an email platform that can hold a few fields alongside each address.
Most small businesses already hold this and have never used it for anything except accounting, which is why the reminder message is available to almost everyone and sent by almost nobody.
Frequently asked questions
How should a small business segment its email list?
By what work each person had done, when they had it, and whether they are a customer at all. Those three splits produce almost all the available relevance.
Why does standard segmentation advice not apply?
It assumes a list large enough that dividing it leaves useful groups, and behavioural data from frequent purchases. A service business has neither.
What is the highest-performing segment?
Customers whose work is due, receiving a reminder tied to their own date. It is relevant to everyone who gets it, which no message to the whole list can be.
Should quoted enquiries be kept separate from customers?
Yes. They are a different audience needing a different tone, and the basis for contacting a past customer differs from the basis for contacting an enquiry that never proceeded.
What segmentation should I skip?
Engagement scoring, frequency modelling, split testing on groups too small to interpret, and personalisation beyond a first name when the message is already specific about the work.
Where does the segmentation data come from?
Your invoicing system. Export the address, the date, and what work was done. Most small businesses already hold this and have only ever used it for accounting.
West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.
List of past customers you have never divided up?
We turn your job records into reminder triggers, so each person gets a message about their own work rather than a mailshot.
Start a Conversation