How repeat rate is calculated
Divide the number of customers who have placed more than one order by the total number of customers, within a stated period. Simple enough, but the way it is usually reported hides a problem: measured across all time, it flatters you, because customers acquired years ago have had years to come back while last month’s have had weeks.
The version worth trusting is measured by cohort. Take everyone acquired in a given month, wait a fixed window, and count how many of them bought again inside it. Now every cohort is being judged on the same terms, trends are visible, and a change in the product or the service shows up as a change in the number rather than being absorbed by the older customers.
It is not the same as how often people buy. Repeat rate counts heads — did this person return at all. Frequency counts orders. You want to know both, because a strong frequency built on a weak repeat rate means a small group is carrying the business.
Why repeat rate matters
It is the earliest honest verdict on whether the thing you sell is any good. Advertising can persuade someone to buy once. Nothing persuades them to buy again except the product, the delivery, the service and the price being what they expected. A first order says your marketing worked; a second says your business did.
It also moves the numbers everything else depends on. Customers who return raise lifetime value, which lifts the acquisition cost you can afford, which decides whether you can compete in a paid auction at all. Businesses that appear to have an advertising problem often have a repeat purchase problem showing up one step downstream.
Common mistakes with repeat rate
Measuring across all time is the first, and it produces a figure that improves automatically as the business ages regardless of whether anything got better. The second is failing to set a window, so different reports quietly measure different things.
The third is broken identity. Guest checkouts, a second email address, a spouse’s phone number or an order taken over the phone all create a new customer record for somebody who is in fact returning, and the rate reads low for no real reason. The fourth is reading a young cohort too early and treating an incomplete answer as a bad one.
How to act on it
Measure by acquisition cohort with a fixed window, and fix the identity problem before you trust the number — matching on email and phone at checkout is unglamorous work that changes the figure more than most campaigns do.
Then treat a weak rate as a business question rather than a marketing one. Ask lapsed customers directly why they did not come back; the answers are usually specific and fixable, and they are cheaper to collect than to guess. Improve the post-purchase experience — delivery, onboarding, the first support interaction — before adding budget, because acquiring more customers into a leaking business simply spends more money to lose more people. When the rate is rising, that is the moment to spend harder on acquisition, not before.
