Loyalty glossary · 7. Data and analytics (31)
Purchase Frequency
Purchase frequency is the rate at which a member transacts with a loyalty programme, expressed as the average number of purchases in a year or the typical interval between purchases. It drives point accrual, activity-based qualification, and the assumptions in actuarial models.
The average purchase frequency in a loyalty programme is a statistical artefact that hides more than it reveals. A few members who buy every week pull the mean upward, while the majority buy once or twice a year and go quiet. Reporting a single number makes the base look healthier and more homogeneous than it is.
Operators who optimise for purchase frequency usually end up chasing short intervals with discount-led tactics. That lifts transactions without lifting loyalty, and it trains members to wait for the next offer. The behaviour that improves the metric is often the cheapest behaviour to buy and the least durable to keep.
Work the effect of a frequency assumption, because the gap is larger than it looks. A member who buys every 2 months and earns 200 points per purchase accrues 1,200 points in 12 months. Stretch that interval to 3 months and the same member accrues 800 points, a drop of 400 points per year. If the programme had forecast 1 billion points of accrual at the 2 month pace, the 3 month pace delivers only 667 million points, a 333 million point miss. At 1 cent per point, that is 3.33 million dollars less liability than planned.
Activity-based qualification turns frequency into a gate. A tier that demands 12 stays a year treats a member who makes 11 high-value stays as less valuable than one who makes 12 low-value stays, which is nonsense. Actuarial models that assume a fixed purchase interval for a cohort will misprice the programme when the mix of members changes, because frequency is not a stable parameter.
Purchase frequency belongs in a dashboard, not in a strategy. The useful version is a distribution by cohort, not a single average, and it should be read next to recency and monetary value. A programme that manages only purchase frequency will find itself optimising for the most promiscuous members and ignoring the most profitable ones.