Behavioral Segmentation: Definition, Variables & Examples
Seven behavioral variables, two brand examples and coffee-subscription rules for choosing an offer and checking later orders.
Behavioral segmentation groups customers by actions such as how recently they bought, how often they order and how they use a product. Use those patterns to find groups worth investigating. Then ask what each group needs before choosing an offer, and measure later purchases to check whether the action worked.
Behavioral segmentation variables
Start with a few actions that could change the offer. Purchase records can establish recency or frequency; benefits sought need a question about what the buyer wanted. Add demographic segmentation when it helps explain who the action-based groups cover.
Variable | What it measures and example |
|---|---|
Purchase behavior | Patterns in choosing and ordering. Example: First-time buyers compared with repeat purchasers. |
Recency | Time since the last completed purchase. Example: Coffee buyers with no completed order in the past 90 days. |
Frequency | Number of purchases in a defined period. Example: One order compared with three or more orders in 90 days. |
Usage rate | How much or how often the product is used. Example: Daily coffee drinkers compared with occasional drinkers. |
Benefits sought | The outcome the buyer wants; ask about it. Example: Convenience compared with exploring different roasts. |
Loyalty | Repeat choice and use of alternatives over time. Example: Regular repeat buyers who also buy competing coffee. |
Occasion | The event or situation associated with use. Example: Weekday refills compared with coffee bought for guests. |
Behavioral segmentation examples: Starbucks and Spotify
Starbucks Rewards connects qualifying purchases with Stars and rewards. A loyalty study can distinguish purchase frequency from reward redemption: earning or redeeming a reward is an action, while exclusive loyalty to one café needs more information.
Spotify Wrapped summarizes listening activity, including top songs and genres. This illustrates how usage patterns can shape the content someone sees. For a research brief, ask which listening patterns call for different discovery examples; listening volume alone does not explain why someone listens.
These public features illustrate behavior-based experiences. The proposed research questions are ours; the features do not reveal either company’s internal segment rules or prove a commercial effect.
Book a demo to compare feasible offers for groups defined by behavior and buying situations. See our comparisons with human research and keep recorded purchases separate from simulated responses when choosing an action.
Example: choosing a coffee-subscription offer
In this hypothetical coffee-subscription example, the team wants to encourage another order. Someone who has not bought for 90 days might still have coffee, need a different delivery schedule or have disliked the roast. Each explanation calls for a different response.
Before building a complicated set of segments, identify the decisions your team can actually make. For the coffee subscription, you could change the quantity, timing, product or price of the next offer.
Customer situation | What to ask and what to try |
|---|---|
They still have coffee at home | Ask how much remains and when they will need more. Try a smaller pack or a later refill. |
Deliveries no longer fit their routine | Ask what changed in their schedule. Offer more control over delivery timing. |
They disliked the last roast | Ask what disappointed them about the taste. Offer product support or a different roast. |
The next order feels too expensive | Ask whether price is the reason they are delaying. Consider a different price or bundle. |
If price is the obstacle, test whether a discount changes the next order. If someone already has enough coffee, a smaller pack or later delivery is another option to compare. An order pattern should start an investigation, rather than stand in for an explanation.
Define the groups so every account has one place
Save the campaign’s eligible accounts and completed orders as of a stated date. Set the time, time zone and included date range before grouping people. Exclude canceled purchases, decide how refunds count and check whether guest orders belong to an existing customer.
For this example, assign each account using the rules below, in order. Once an account matches a rule, stop; do not put it in another group.
- Unknown: you cannot reliably connect the account with its purchase history. Check the records before treating the customer as new or lapsed.
- New: the history is reliable and the first-ever completed order was within the past 30 days. Focus your questions on the first delivery and what the customer expects next.
- Repeat: among the remaining accounts, there were at least three completed orders in the past 90 days. Find out whether the customer values a predictable delivery plan, flexibility or savings.
- Lapsed: among the remaining accounts, there was a completed purchase, but none in the past 90 days. Ask what changed after the last order.
- Other: the history is reliable, but the account meets none of those rules. Keep it here while you decide whether a different action is useful.
For example, a customer whose first order was 20 days ago and who has already placed three orders is New. New comes before Repeat, so the account is counted once. A customer with one completed order 45 days ago is Other: they are beyond the New window, below the Repeat threshold and have ordered too recently to be Lapsed.
The order is Unknown → New → Repeat → Lapsed → Other. Unknown comes first because missing purchase records can make a regular buyer look new or lapsed.
Use the buying cycle to choose your time windows. Ninety days may be a long absence for a monthly coffee order and perfectly normal for a product bought twice a year. The 30-day and 90-day cutoffs here are example rules to adapt to your category.
Start with the last purchase
Once you know the group, ask about the latest purchase before showing an offer:
- “Think about your last order. What happened afterward?”
- “What, if anything, would make another order useful?”
For a lapsed coffee customer, follow up on how much coffee remains, when another delivery would become useful and what they have chosen instead. Leave room for a reason your team has not considered. Someone may have moved, changed their routine or stopped drinking coffee.
Ask these questions before showing a discount or a new bundle. Otherwise, you risk getting reactions to your proposed solution without understanding the problem it needs to solve. Our segmentation guide explains how to connect customer needs with the action each group should receive.
The explanation determines the offer
Suppose the customer still has plenty of coffee. Compare a smaller pack or a later refill. If their routine has changed, compare scheduling flexibility. If the roast disappointed them, investigate the experience before asking them to buy more of it.
Keep the commercial terms visible when you compare offers. A smaller pack with a different unit price and a delayed delivery with the same quantity solve different problems. Your study should tell the team which tradeoff matters to the customer and why.
Purchase history tells you whom to contact and which purchase to ask about. The explanation tells you which offer to develop. Several lapsed customers can have different needs without requiring a new segment for every answer.
Start with the few distinctions that change the action. Each extra group creates more campaign versions, rules and operational work. Compare the targeted approach with a simpler shared offer before committing to that complexity.
Record quality determines the assignment
Before activation, check a few accounts near each cutoff and a few with missing or duplicated records. These checks answer practical questions:
- Does a customer with three orders since their first purchase 20 days ago appear only in New?
- Does a known customer with one order 45 days ago remain in Other?
- Does a customer with no completed order in the past 90 days enter Lapsed only after their earlier purchase history has been confirmed?
- Do unmatched guest orders or duplicated accounts create a false lapse?
Report Unknown and Other too. Unknown shows how much purchase history needs checking. Other shows how many eligible customers fall outside the chosen rules. Do not drop either group from the report.
Then check whether you can reach the selected customers and deliver the proposed offer. A group is only worth targeting if your systems can find it and your team can act on it.
Judge the action by what happens next
Save the original groups, date and eligible customer list before sending offers. Keep those assignments fixed for the evaluation, even if customers later change groups. Count completed orders in a new period after that date. Past purchases used to qualify for Repeat cannot show that the new campaign worked.
For the coffee subscription, your report should show which offer each group received, how many eligible customers were included and how many completed another order during the follow-up period. Keep the purchase definition and follow-up period consistent across the groups you compare.
Within each saved group, randomly assign eligible accounts to the targeted offer or a stated shared-offer/control condition. Compare completed orders over the same follow-up period. Orders after a campaign alone do not tell you what it caused.
Use the result to decide whether the extra segmentation is worth maintaining. If a simpler offer serves customers just as well, keep the simpler system. If scheduling flexibility solves a clear problem for one group, that difference gives you a reason to keep it.
From a lapsed coffee account to an offer comparison
A Sapien study can compare simulated responses to a smaller pack, a later refill and scheduling flexibility, then explore the reasons across your declared coffee-buyer groups. Supply complete prices, quantities and delivery terms with the group rules and category buying cycle.
Keep simulated responses distinct from transaction records. Your customers' later completed orders remain the reference for actual repeat purchasing.
Use the coffee offer brief with Sapien to decide which stock, timing, roast or price explanations the study must distinguish. For background on usage and buying occasion, see OpenStax’s segmentation chapter.
