Repeat purchase rate:
formula and worked example

What share of your buyers come back

Repeat purchase rate is the proportion of customers in a defined group who have bought from you more than once. It separates a business that keeps serving the same people from one that has to pay to find a new buyer for every sale. For a web shop, a subscription box or a neighbourhood service, it is one of the earliest signals of whether the first order left the customer happy enough to return.

The metric records behaviour, not satisfaction directly. A high figure can come from a product people simply have to restock, such as groceries or consumables, while a modest figure can be perfectly healthy for a wedding photographer or a sofa seller. The number becomes meaningful only when you compare it with your own earlier periods and with businesses that sell on a similar buying cycle.

It also changes how you read growth. Two shops can both double their order count in a year, one by adding a flood of first-time buyers and the other by getting existing buyers to order again. The second is usually cheaper to sustain, because a returning customer needs no acquisition spend, no trust-building and often less support. Repeat purchase rate is the figure that tells those two stories apart.

The formula and its two inputs

• Repeat customers: the number of distinct people with two or more orders inside the window you chose. Count people, not orders.
• Total customers: the number of distinct people who bought at least once in the same group and window.
• Window: not a symbol in the formula, but it decides the answer. A twelve-month window will always show a higher rate than a three-month one.

Both numbers must describe the same group. If the denominator is everyone who has ever bought and the numerator counts only this quarter's returners, the ratio mixes two populations and says very little.

Worked example: 180 returning customers out of 900

An online store looks at everyone who made a first purchase during the last twelve months. In total 900 distinct customers bought, and 180 of them placed at least one further order.

1. 1Count distinct customers in the group: 900.
2. 2Count how many of them ordered more than once: 180.
3. 3Divide: 180 ÷ 900 = 0.20.
4. 4Multiply by 100 to express it as a percentage: 20%.

One customer in five came back. The other 720 bought once and, so far, have not returned, which is the pool a win-back email or a better first-order experience is aimed at.

How to read a 20% repeat rate

There is no universal good number. Conventions differ sharply by category, so read 20% against the natural reorder rhythm of what you sell.

• Fast-moving or consumable goods usually produce higher rates, and a figure near 20% for such a product often points to a retention gap.
• Durable or occasional purchases, such as furniture or appliances, often sit in single digits without anything being wrong.
• A rising trend across comparable cohorts matters more than any single reading.
• A rate that climbs while total customers shrink can simply mean the one-time buyers stopped arriving.

It also helps to pair the rate with how long returners took to come back. A customer who returns in week three and one who returns in month eleven are not the same signal.

One more habit helps: look at the gap between first and second order. If most second orders arrive within a few weeks, a reminder or a replenishment prompt timed at that point can pull more customers through. If the second order tends to come after many months, the more useful lever is staying visible, through email, content or an occasional offer, so you are still remembered when the need returns.

What happens when repeaters or the base change

With 900 customers in the group, every 9 additional repeaters adds exactly one percentage point. The reverse also bites: if the base grows to 1,000 through a heavy acquisition month while repeaters stay at 180, the rate drops to 18% even though nothing got worse for existing customers. Always ask whether a change came from the numerator or the denominator before drawing conclusions.

Mistakes that distort the rate

• Counting orders instead of customers, so one person with five orders inflates the numerator five times.
• Comparing a young cohort with an old one. Customers acquired last month have not had time to return, so their rate looks artificially low.
• Treating one person with two email addresses or a guest checkout as two separate customers.
• Leaving cancelled, fully refunded or fraudulent orders in the count, which can turn a non-purchase into a repeat purchase.

A practical safeguard is to freeze a definition document before comparing anything. Write down how a customer is identified, which order statuses count, which channels are included and how the window is cut. When a new analyst or tool appears, the same document lets them reproduce last quarter's figure first, and only then compute this quarter's. Most disagreements over the repeat rate turn out to be disagreements over definitions, and they are settled in minutes once those definitions are written down rather than remembered.

A quick check before you act on it

Suppose you ran a discount campaign in March and your repeat rate for March buyers came out at 11% against 20% for the rest of the year. That does not prove the campaign was bad. Discount-led buyers often have a different intent, and the cohort may simply be too young. Wait one full reorder cycle, then compare like with like.

When reporting the figure to partners or a team, state the cohort, the window and the definition of a purchase next to the percentage. A line such as 'customers who first bought in January to March, measured to the end of September, excluding cancelled orders' costs a dozen words and prevents most of the arguments that follow a surprising number.