How to measure repeat customer rate (and what to do with it)

Repeat customer rate is the most commonly quoted retention metric and the most commonly miscalculated one, usually because the measurement window is chosen arbitrarily. Getting the window right matters more than getting the formula right.

How do you calculate repeat customer rate?

Divide the number of customers in the period who had bought from you before by the total number of customers in that period, and express it as a percentage. A shop serving 400 distinct customers in a month, 120 of whom had bought before, has a repeat customer rate of 30 per cent.

The word doing the work is "distinct". Counting transactions rather than customers inflates the figure for any business where regulars buy several times a week, which is most of the ones that need the metric.

What period should you measure over?

Use a window of roughly three purchase cycles for your business, which means a month for a café and a year or more for a garage. A window shorter than one cycle mechanically reports almost everyone as a new customer.

This is where most reported figures go wrong. A salon measuring its repeat rate monthly will report something near zero and conclude it has a retention crisis, when in fact its clients are simply on an eight-week cycle.

What is a good repeat customer rate?

It depends so heavily on the business that cross-industry benchmarks are close to useless: a café with a 30 per cent repeat rate and a jeweller with a 30 per cent repeat rate are in completely different situations. The comparison worth making is against your own figure last period.

Two things make published benchmarks misleading. They rarely state the window, which changes the number by a factor of several, and they rarely state whether they count customers or transactions. A figure without both is not comparable to anything, including itself a year later.

Why is repeat customer rate not enough?

Because it is a ratio, and a ratio can improve for bad reasons. A business that loses half its new customers will see its repeat rate rise, because the denominator shrank: the number went up and the business got smaller.

It is also a lagging figure that tells you nothing about who to contact. A rate of 28 per cent is a fact about last quarter; it does not name a single person, and no action follows from it.

Which numbers are more useful?

Two: the lapsed-customer list, and the repeat purchase interval. The first is a set of named people who used to buy from you and have stopped, and the second tells you when to consider someone lapsed in the first place.

• Lapsed customers: regulars who have passed two to three times their normal gap without a visit. This is a list, not a number, and every name on it is a phone call or a message.
• Repeat purchase interval: the median gap between one customer's visits. It sets the lapse threshold and the reminder timing for every other metric.
• Enrolment rate: the share of transactions that result in a loyalty enrolment, for as long as the programme is new.
• Reward completion rate: the share of enrolled customers who reach the reward at least once. A low figure means the threshold is too high.

The lapsed list is the one to act on weekly. Repeat customer rate is worth calculating quarterly to see a trend, and it should never be the number a decision is made on.

How do you measure any of this without a loyalty programme?

For a cash-and-card business, you generally cannot. A card payment can sometimes be matched to a returning customer through a payment processor's tools, but cash visits are invisible and most small tills do not identify customers at all.

This is the quiet reason a loyalty programme is worth running even where the reward changes little. The programme's real output is a customer record: who came, how often, and when they stopped. Loonine's calculator on the homepage runs a conservative retention model on those figures, and it shows its assumptions rather than hiding them.

What is a modest improvement worth?

Enough to matter and much less than the frequently repeated claim that a five per cent retention increase raises profit by 25 to 95 per cent. That figure comes from a specific study of contractual businesses and does not transfer cleanly to a café.

A defensible way to size it for your own business: take your number of active customers, your average spend per visit, and one extra visit per customer per year. For 300 customers at €8, that is €2,400, a real figure you can check, arrived at without borrowing anyone's percentage.

How do you spot a lapsed customer?

Take each customer's own normal gap between visits and treat two to three times that gap as lapsed. A café regular who visits every three days is a concern at three weeks; a salon client on an eight-week cycle is not a concern until four months.

A single threshold across the whole customer base is what makes most lapsed reports useless. The same ninety-day rule flags a café's best customer far too late and a garage's perfectly normal one far too early.

What is a cohort, and is it worth the effort?

A cohort is the group of customers who first bought in the same month, tracked forward. It is worth the effort because it separates a retention problem from a recruitment one, which a single repeat rate cannot do.

The minimum useful version is one line per month: of the customers who first appeared in March, how many came back at all within ninety days. If that figure is steady while your overall repeat rate falls, you are recruiting more customers rather than keeping fewer, which calls for a completely different response.

Which figures should you ignore?

Total enrolments, total stamps issued, and any percentage borrowed from another industry. All three move in the right direction whatever happens and none of them names a customer or implies an action.

Total enrolments is the most misleading because it only ever goes up. A programme that stopped working six months ago still shows a growing enrolment total, which is why enrolment rate, enrolments as a share of transactions, is the version worth putting on a wall.