Most dispensary SMS programs run on weak segmentation: "everyone who bought in the last 90 days," "VIPs," "the loyalty list." Those segments are too broad to drive real lift. The operators producing the strongest results are the ones running 6-10 working segments at any given time, each one tied to a specific behavioral signal.
1. Category affinity is the foundation
Customers buy in patterns. A customer whose basket history is 80% flower is a different customer than one whose history is 80% concentrates — and they should not get the same promotional content. Build category-affinity segments off live POS data: flower-led, concentrate-led, edible-led, pre-roll-led, mixed. Run separate campaigns into each.
2. Visit cadence segments power retention
High-frequency visitors (weekly+) need different cadence and content than monthly visitors. Lapsing visitors (gone past their normal interval) need win-back. New customers (under 30 days) need welcome and onboarding. These four cadence segments alone — high, monthly, lapsing, new — typically cover 80% of revenue lift from cadence-aware segmentation.
The mechanic
Cadence segments should be derived from per-customer baselines, not population averages. A 21-day gap is lapse risk for a normally-weekly customer; it's nothing for a normally-monthly one.
3. Basket size separates VIPs from volume buyers
Two customers who visit at the same cadence can have very different LTV based on basket size. Layer basket-size into your segmentation so high-basket customers get higher-value offers and personalized touches; volume buyers get cadence-led promotional flow.
4. Tier mechanics from loyalty add a fourth dimension
Loyalty tier (VIP, gold, silver, base) is an explicit signal of program engagement and lifetime value. It belongs in your segmentation alongside category, cadence, and basket. Top-tier customers should never get the same template as base-tier — period.
5. Lapse risk is the highest-leverage trigger
A customer slipping past their normal visit interval is at risk of becoming lapsed. The window to catch them is short — usually 1-2x their normal cadence. After that, win-back rates drop off a cliff. Build lapse-risk scoring and trigger win-back automations the same day a customer crosses the threshold.
6. Layer segments, don't multiply them
The mistake operators make: building 30 segments and then sending to them all. The right approach is to build the core dimensions (category, cadence, basket, tier, lapse risk) and let the platform compose them on the fly for each campaign. You're not picking from a menu of 30 audiences — you're describing the audience you want for this campaign and the platform composes it from the dimensions.
7. Measure lift per segment, not gross response
A "high-response" segment that would have visited anyway has zero incremental lift. Build hold-out groups into segment-level reporting so you see incremental revenue per segment, not just gross attribution. The segments that produce real lift are the ones you keep investing in.