Is your store growing through new or returning customers?

When revenue grows, it is tempting to take it as proof that the store is doing things right. Yet the same growth can come from three different places: the store is winning more new customers, past customers are coming back more often, or revenue is simply carried by a base the store built in the past. Each path leads to a different next decision. So split growth into new and returning customers – and do not read only their share of revenue. That share moves even when customer behaviour has not changed at all.

Break growth into four numbers

Revenue for a period can be written as the sum of two parts:

• new customers × average revenue per new customer,
• returning customers × average revenue per returning customer.

The number of returning customers breaks down further: how many past customers the store has and what share of them bought again in the period. We will call that share the return rate. Research on valuing e-commerce firms models the value of a customer base in the same way – acquisition, ordering and spend per order separately.

An illustrative example with invented numbers, revenue excluding VAT:

Revenue grew by 24% and the returning share rose from a third to a half. At first sight it looks like a loyalty success. But the return rate did not change, and neither did spend per customer. All of the growth came from the base of past customers doubling – while new customers fell by a tenth. The constraint in this store is not retention but acquisition. Once the base stops growing, revenue from returning customers will slow as well.

Why the returning share misleads

The base grows on its own. Every new customer adds to the pool of people who can come back. In a young store nearly all customers are new; in an older one returning customers naturally increase even if nobody behaves differently.

A store cannot see who has left. Unlike a subscriber, an online shopper does not announce that they are done. The store therefore cannot reliably tell a customer who has left from one who is simply between purchases. Customer-base researchers use Amazon as an example: defining an “active customer” as one who ordered in the past 12 months is an arbitrary cut-off; with nine months the base would look smaller although its true size would be unchanged. How many customers count as “returning” thus also depends on the chosen window.

Older customers look more loyal. The customers least inclined to come back drop out of every group first. Those who remain return more often, so a group’s retention rate typically rises over time – not because individual customers become more loyal, but through sorting. The study comes from a subscription setting, but the principle holds for an online store too: comparing the return rate of an old base with that of fresh customers compares different groups, not different loyalty.

Growth usually comes from more buyers, not more loyal ones

Brand research adds a broader observation. Smaller brands have far fewer buyers in a given period, and those buyers also buy them less often. Loyalty moves together with size rather than independently of it. When brands grew, consumer panel data from the UK and the US showed buyer numbers rising, on average, far more than loyalty, with the largest gain among occasional buyers. An analysis of 24 datasets across 17 categories in China, Malaysia and Indonesia reached the same conclusion: share growth came with a larger change in the number of buyers than in any other metric.

These studies concern packaged-goods brands, not online stores, and cannot be read as a law for your business. As a principle, though, they are useful: a plan that expects most growth from existing customers buying much more relies on the figure that moves least in the data. That does not mean retention is unimportant. A store whose new customers disappear after their first purchase pays for every order again. It simply means retention rarely carries growth on its own.

Align definitions before you compare

Who counts as one customer

Counting customers by registered accounts is the most convenient and least reliable option. Shoptet’s customer list does not show people who bought without registering, and it links orders to an account by e-mail. Someone who repeatedly buys as a guest would look new every time if you counted accounts. The order e-mail, converted to lower case and stripped of spaces, is more reliable. It is not perfect either: a person with two addresses is counted twice. Record that imprecision rather than estimating around it.

When your history starts

If your data starts in January, a customer who last bought in December looks new in January. The first months of history therefore overstate new customers. Compare only periods with enough history behind them – at least as long as it usually takes your customers to come back.

A new user in Google Analytics is not a new customer

GA4 counts a “new user” from the first visit to the site and a “returning user” from a previous session. On the web it relies on an analytics cookie in the browser; if a visitor declines analytics cookies, Google tags neither read nor write them. In the EU you need that consent for anything that is not technically necessary. A loyal customer on a new phone is therefore a new user to GA4. Use orders for new and returning customers and keep Google Analytics for traffic trends.

What exactly “returning” means

Choose one definition and stick to it: a returning customer is one whose order in the period is not their first order in the entire history. Also decide whether you count customers or orders – a customer who places both a first and a second order in the same period is otherwise new and returning at once. Exclude cancelled and returned orders the same way in both periods.

Measure loyalty on comparable cohorts

To find out whether customers return better than before, do not compare shares of revenue; compare groups of customers by the month of their first purchase. For each group, calculate what share bought a second time within a chosen number of days, and compare only groups for which that full window has already passed. Set the window from your own purchase rhythm rather than a general rule. And compare the same season: customers from the Christmas weeks behave differently from summer customers.

Which constraint the breakdown points to

Treat visits from new people in Google Analytics as a trend rather than an exact count, because of cookies and consent. And if the return rate falls after a large discount campaign, first check whether the mix of new customers has changed. A different group of people is not the same as worse customer care.

What to do now

1. Align the definition of a customer. Count by order e-mail, not by registrations, and note when your history starts.
2. Calculate four numbers for two comparable periods. New and returning customers and revenue per customer in each group, excluding VAT and with cancellations treated the same way.
3. Add the return rate. How many past customers bought again in the period, and how cohorts of the same age return.
4. Choose one constraint from the table and check it with one more number, such as visits from new people or conversion rate.
5. Write down a hypothesis with a review date. What should change, in which number and by when.