Purchase frequency is total orders / unique customers transacting over a defined period. The denominator deserves a label: some operators divide by every customer in the file (including non-buyers in the period), which produces a smaller number; the more common DTC convention is to divide by active buyers in the window. Both are defensible, but a dashboard that switches between them silently is comparing different metrics.
Period choice matters more than the headline figure. A trailing-12-month frequency of 2.1 on a brand with a ~9-month average inter-purchase interval reads very differently from the same 2.1 on a brand with a ~3-month interval — the first is a healthy second order arriving on cycle, the second is a customer base buying half as often as the category supports. The metric does not normalize against the natural repeat cycle, so cross-category comparisons of the raw number are misleading unless the cycle is held constant.
A period frequency is also a population snapshot, not a per-customer trajectory. The same 2.1 can come from a uniform “everyone buys twice” pattern or a heavy-tailed “most buy once, a loyal 10% buy a dozen times” pattern, and the operator response to those two distributions is opposite. Cohort frequency by acquisition month tells a more honest story — it isolates how a single acquisition group is maturing instead of averaging across mixed-tenure buyers.
Purchase frequency is the third leg of the LTV identity, sitting alongside AOV and gross margin. It is what separates LTV growth driven by raising prices from LTV growth driven by deepening the relationship — frequency only moves when customers come back more often. It is also not the same metric as repeat purchase rate: repeat rate is the share of customers who buy a second time at all (binary), and frequency is the mean order count across the population (a magnitude). A high repeat rate with a low frequency (most repeaters come back once and stop) reads very differently from a low repeat rate with a high frequency (a small loyal core orders constantly). Reading the two together is the move.
Of the retention metrics, frequency is the most directly intervenable: subscription conversion, replenishment reminders, post-purchase email and SMS flows, loyalty programs, and category expansion all tend to show up as frequency lift before they show up as LTV lift, because the LTV term lags by customer lifespan. Use trailing-12-month frequency for cross-brand benchmarking, since the window absorbs seasonality and most natural repeat cycles; use cohort frequency by acquisition month to diagnose whether recent retention work is paying off, since the cohort view isolates the change from base-rate drift.