What is a good repeat purchase rate?

Before judging the rate, work out how often a satisfied customer would naturally need to buy again. That interval decides what is possible.

A consumable that runs out every month can produce many orders a year from a happy customer. A durable item bought once every few years cannot, no matter how good the marketing. Comparing the two is meaningless, and comparing either to a general benchmark is worse.

So the first task is to find your own cycle. Take customers who have ordered at least twice and look at the gap between their first and second orders. Not the average, the distribution, because most categories show a cluster where the product runs out and a long tail of people who came back for another reason.

That cluster is your window. If it sits at around eight weeks, then a customer who has not returned after four months has probably gone, and your measurement period should be built around that fact.

For durable goods the second order is a different product, not a repeat of the first. A customer who bought a jacket may buy a bag. The cycle is set by the season or the occasion rather than by consumption, and the rate you can expect is lower and slower.

Once you know the cycle, the question stops being what is good and becomes what share of customers come back inside the window when they had every reason to.

Three different numbers all get called the repeat rate, and they answer different questions.

Repeat customer rate is the share of customers who have ever placed more than one order. It is simple and it drifts upward over time, because customers only ever move from one order to two, never back. On a growing store it is diluted by new customers who have not had time to return yet.

Orders per customer is total orders divided by total customers over a period. It reacts faster and is harder to game, but a handful of very frequent buyers can carry it.

Repeat rate within a window is the cleanest for decisions. Of the customers who first bought in a given month, what share bought again within your reorder cycle. It answers a question you can act on, because you know which customers, when, and what you sent them.

Pick one, define it in writing, and use it consistently. Most disagreements about whether retention is improving turn out to be two people using two definitions.

Whichever you pick, separate subscription orders from one off orders. Subscriptions produce repeat orders by design, and mixing them in makes the acquisition side of the business look healthier than it is.

A single store-wide figure moves for reasons that have nothing to do with retention. Grow fast and the rate falls, because the denominator fills with people who have not had time to come back. Stop advertising and it rises, because the base ages. Neither tells you whether customers are happier.

Cohorts fix this. Group customers by the month of their first order, then track what share of each group has bought again after thirty, sixty and ninety days. Now you are comparing like with like, and you can see whether the customers you acquired in March behave better than the ones from January.

The pattern reveals things a blended number cannot. A cohort acquired during a heavy discount period usually returns worse, because the offer attracted deal seekers rather than customers. A cohort acquired from a specific channel may return far better, which changes what you are willing to pay for that channel.

Read the curve rather than a single point. Most of the second orders arrive early, and the shape of that curve tells you when your messages should land.

This is also the honest way to test a retention change. Introduce a new post purchase flow, then compare the cohorts before and after at the same age. Comparing an old cohort at twelve months with a new one at two months will show you a decline that does not exist.

Four causes cover most of it, and they need different responses.

The product did not deliver. The most common reason a customer does not come back is that the thing was not as good as expected, arrived late, arrived damaged, or was harder to use than the page implied. No email sequence fixes this. Read your reviews and your support tickets first, because they contain the answer.

There was no reason to return. Some catalogues have no natural second purchase. If you sell one thing and it lasts, the second order has to be created: a companion product, a larger size, a gift for someone else, a refill.

The timing was wrong. Messages sent on a calendar rather than on the product's cycle arrive after the customer has already bought elsewhere. If the product lasts six weeks and the reminder goes out at twelve, you are reminding people who already replaced you.

The customer was bought badly. Heavy discounting, deal sites and a certain kind of cheap traffic bring customers who buy the offer, not the brand. Look at repeat rate by acquisition channel and you will often find one channel producing customers who never return, which changes what that channel is really worth.

Work through them in that order. The first two are business problems. The last two are marketing problems, and marketing cannot fix the first two.