Academy › Publishing › Analytics

Out of every ten, how many come back

By the end, you'll know your real repeat customer rate, calculated from your own data for a specific period, and the calculation mistake that skews it the most.

By Yodai · Published September 6, 2026 · Updated October 12, 2026

For this calculation, you need your rule for what counts as a customer coming back, written in one sentence. But if someone asked you right now what percentage of your customers repeat, you'd probably answer with a feeling instead of a number: "a lot," "not many," "about half."

That feeling won't help you decide anything. It doesn't tell you whether you're doing better or worse than last quarter, or whether it's worth spending on a message to bring back the customers who've drifted away. Once you've set your definition, that percentage comes from a three-step calculation, and it's easy for your gut to mislead you until you run the numbers yourself.

The calculation, in three steps

Step three is the one people get wrong most often. You're not counting how many visits each customer had within the period: you're counting whether, before the period started, that customer already met your definition of a repeat customer. Someone who bought for the first time on day one of the quarter and came back in week six doesn't count as a repeat here: she's a new customer who happened to make a second visit within that same quarter. That's a different number, useful for something else, but it's not the period's repeat percentage.

Take a hair salon that served forty customers last quarter. Of those forty, fourteen had already been in before the quarter started. Fourteen divided by forty is 0.35. Times 100, and 35 percent of the quarter's customers were repeat customers.

That thirty-five percent depends entirely on which list you use for "your customers." If instead of the quarter's list you use the list of every client who's ever bought something since the salon opened, the percentage changes a lot, and not because the business changed one bit.

The same math works for a dental practice. Last year it saw 180 patients. Of those, 61 had already been patients before the year started. 61 divided by 180 is 34%, almost the same rate as the salon, even though a dental patient's visit cycle and a salon client's don't look anything alike. The repeat rate doesn't tell you how often someone comes back: it only tells you what share of the people who came through the business had already been there before.

It works the same way for an online accessories store. Over twelve months it sold to 500 different customers; 95 of them had already bought before that year started. That's 19%, quite a bit lower than the salon or the dental practice. That doesn't necessarily mean it's doing worse: a store that ships nationwide can have a lot of one-time customers buying a gift or a one-off item, and its rate doesn't compare well to a neighborhood salon's. What matters is comparing it to itself, period over period.

The mistake that skews the number

Now you have your real repeat customer percentage, calculated the same way every time, from your own numbers instead of a gut feeling. What's left is figuring out how often a repeat customer actually comes back, so you don't mistake someone who's simply running a bit late for someone who's gone for good.

With that percentage, you already know how many come back. What you don't know is how often the ones who return actually do, and that interval, average or median depending on how much data you have, is what you calculate now, service by service.