A segment earns its place when it changes at least one decision: the promise, timing, content, or call to action.
Start with intent, not columns
It is tempting to create a group for every field you own. Start with the campaign instead: who genuinely benefits from this offer or information? Then identify the data that can find those people.
Demographics are not automatically useful. Recent behavior, the relationship with the business, and a stated preference often create groups that lead to a clearer action.
- Journey stage: prospect, new customer, regular, or inactive.
- Interest: product, service, or content category.
- Recency: last purchase or interaction.
- Context: city, store, event, or preferred language.
Begin with one or two criteria
Over-segmentation creates tiny groups that are hard to maintain and impossible to compare. Begin with one primary criterion and one piece of context.
For example, select customers who bought a course and have shown no activity for six months. That combination supports a specific re-engagement message without requiring a complicated scoring model.
Use names the whole team understands
Name groups by their rule. “Footwear customers · purchase under 90 days” is more useful than “Final group 3.” Add a short description when membership is not obvious.
Also state whether a group is static for one campaign or dynamic as contact data changes. Without that distinction, a once-relevant audience may be reused months after it became stale.
Design exclusions from the start
Good targeting defines who must stay out as clearly as who comes in. Exclude opted-out contacts, invalid numbers, people who already converted, and anyone in an incompatible service conversation.
Apply exclusions again immediately before launch. A contact’s status can change between group preparation and the final send.
- Opt-out or withdrawn consent.
- Campaign goal already completed.
- Open support issue.
- Frequency limit reached.
- Essential data missing.
Compare groups, not only messages
Review results by contact group. A message variant can look stronger simply because it reached people who were already more engaged.
Document group size, selection criteria, and exclusions. You can reuse segments that produce useful conversations and remove distinctions that never change the outcome.