Acquisition costs in cannabis are high and the LTV math only works when customers come back. Repeat-visit cadence is the metric that actually predicts revenue — and SMS is the channel that moves it most directly. Done well, the right text at the right moment can compress a 35-day return interval into 21 days. Done badly, the same list burns out in months.
1. Model the real cadence by category
Different product categories have different consumption rhythms. Flower customers typically return faster than concentrate customers. Pre-roll buyers come in more often than edible-only buyers. Model each category's natural cadence from your own POS data and trigger based on those rhythms — not on a flat 30-day rule.
2. Post-visit timing matters more than copy
A thank-you text that fires 4 hours after a visit performs very differently from one that fires 4 days later. Same content, very different result. Test post-visit timing per category (most operators find the sweet spot is 2-6 hours for flower, 24-48 hours for edibles) and let the data set the schedule.
The trap
Most generic SMS tools fire post-visit at a fixed delay regardless of category, customer profile, or product. That's why their lift flatlines. Cadence has to be category-aware.
3. Trigger win-backs on real lapse risk, not arbitrary intervals
A customer who normally visits every 14 days hitting day 21 is lapsing. A customer who normally visits every 60 days hitting day 21 is just on schedule. The same trigger fired on both is wasteful at best and irritating at worst. Build win-back triggers off per-customer cadence baselines, not population averages.
4. Use category affinity to make win-backs specific
Generic "we miss you" texts underperform category-specific ones by wide margins. A customer whose basket history is 80% concentrates gets a concentrate-led win-back; a customer who alternates between flower and pre-rolls gets an offer covering both. The platform should do this automatically from POS data.
5. Reduce burnout by enforcing cadence caps
The fastest way to destroy a repeat-customer program is over-sending. Cadence caps (max N messages per customer per week, regardless of which automation triggered them) protect the relationship. Most operators land at 2-3 messages per week per customer; some VIP programs run higher with content the customer expects.
6. Measure repeat-visit lift against a hold-out
Sending more is not the same as driving more visits. Build hold-out groups into every retention automation and measure incremental repeat-visit rate — not gross visit count. Some flows that look like winners in aggregate are zero-lift when measured properly. Knowing which ones actually move repeat behavior is what compounds.