Time Between Orders
Date: 2026-08-16
The interval that defines what “churned” even means for a non-subscription business. It’s also the timing signal underneath every lifecycle campaign — reach someone at the wrong point in their cycle and the message is noise.
What it is
Time between orders — the interpurchase interval — is the elapsed time from one order to the next, per customer.
It’s a distribution, not a number, and it’s right-skewed like everything else here:
days between order 1 and order 2
▌
▌▖
▌▝▖▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
──┬───┬──────────────────
median mean
120 180
Report the median. The mean is dragged by customers who returned after two years — Mean Median and Mode, Skewed and Heavy-Tailed Distributions.
What it’s for
Three things, and each is load-bearing:
Defining churn. Without a cancellation event, “churned” is a judgement about elapsed time. A workable rule: twice the median interval without an order. At a 120-day median, that’s 240 days — long enough that most genuine repeaters would have returned — Retention and Churn.
Timing lifecycle campaigns. A replenishment prompt landing before someone has run out is ignored; landing after they’ve rebought elsewhere is wasted. The median interval is the anchor — Replenishment Timing, Lifecycle Messaging.
Setting the churn window for winback. Too early and you’re discounting to people who’d have returned anyway; too late and they’re gone — Winback Campaigns.
It shortens with each order
The pattern worth knowing, because it changes campaign timing:
order 1 → 2 median 120 days
order 2 → 3 median 85 days
order 3 → 4 median 60 days
Customers who buy more often buy more often — partly habit, partly that frequent buyers are a self-selected group. Either way, campaign timing should tighten as order count rises, and a single fixed cadence for everyone is wrong at both ends.
Segment it before using it
A blended interval is nearly useless if your range spans categories with different natural cycles:
consumables 45 days
apparel 110 days
homeware 240 days
blended 120 days ← describes nobody
Segment by first category, or by the specific product where replenishment is predictable. Product-level intervals are what make replenishment campaigns work rather than annoy.
Measuring it
- Only customers with two or more orders have an interval. One-time buyers are excluded by construction, so the population is self-selected — Survivorship Bias
- Right-censoring. A customer who bought once six months ago may still return, but has no interval yet. Excluding them biases the median downward, because you’re only counting intervals that have already completed
- Use a fixed observation window and be explicit that recent cohorts are incomplete
- Requires durable identity. Guest checkout across two orders means two customers with no interval between them — Identity Stitching
Where it connects
- Payback Period — when the second order arrives determines when acquisition cost is recovered, so shortening the interval improves cash flow without changing lifetime value at all
- Purchase Frequency — the same information expressed as a rate rather than an interval
- Customer Lifetime Value — the interval sets how many orders fit inside your LTV horizon
- Stockouts and Availability — a stockout at the moment someone would have reordered doesn’t delay the order, it usually loses it
In plain terms: the interval is the clock the whole lifecycle runs on. Almost every retention tactic is a decision about when to act, and this is the number that answers it.