Repeat Purchase Rate: A Practical Guide to Measuring It

Every example and amount on this page is illustrative. It shows the method and is not a benchmark, a quote or advice.

What Repeat Purchase Rate Actually Measures

Repeat purchase rate measures customer behavior: whether customers in a defined group placed more than one order during a defined period. The group is the cohort. It might be customers whose first order occurred in January, or customers who made a purchase during a specific campaign period. The time window might run for 30, 90, or 365 days after that qualifying event.

Both parts must be stated before the number is useful. “January customers” is incomplete unless it means customers who first purchased in January, customers who purchased at any time in January, or another defined population. “Repeat” is also incomplete unless the business specifies which orders qualify and when the observation period ends. A customer who places a second order after the window closes has not repeated within that measurement period; that does not mean the customer will never buy again.

The metric is customer-level and binary for the selected window: each customer either placed at least one additional qualifying order or did not. It therefore counts people, not the number of additional orders. One customer with two extra orders and another with one extra order both qualify as repeat purchasers.

Use the metric to answer: “What share of this customer cohort came back at least once within this period?” Do not use it to answer how much those customers spent, how often they ordered, or how profitable they were. It also does not establish ongoing retention after the window, future purchases, or customer lifetime value. Those require different measurements and, for profitability, cost and return data such as the analysis described in Ecommerce Customer Profitability & Lifetime Revenue.

The Formula in Plain English

The standard formula is: Repeat purchase rate = repeat purchasers ÷ total customers in the cohort × 100. A repeat purchaser is a customer who placed at least two qualifying orders during the measurement window. The denominator is every customer in that same cohort who placed at least one qualifying order. Count customers, not orders: someone with three orders still counts as one repeat purchaser.

The cohort is the defined customer group you are measuring. It might be all customers who placed an order in January, or customers acquired through a specific channel during a quarter. The measurement window is the period in which you look for the additional purchase. Both parts must be stated. “January customers who bought again within 60 days” is a usable definition; “January repeat rate” is incomplete unless the observation period is understood.

A qualifying purchase also needs a clear rule. Many teams count a completed order and exclude cancelled orders. You must decide how to treat returned, refunded, exchanged or partially fulfilled orders, then apply that rule to both the numerator and denominator. A return can still represent a purchase event for a behavioral measure, while a cancelled order may not. The important point is not choosing one universal rule; it is documenting the rule and using it consistently.

Consistency matters because changing the cohort, window or order-status rules changes the result without changing customer behavior. Comparing a 30-day rate with a 90-day rate, or counting cancelled orders in one period but excluding them in another, creates a false trend. Keep the customer definition, qualifying-purchase rule and measurement window fixed when comparing periods. If you change one, label the result as a different calculation rather than treating it as a like-for-like comparison.

A Worked Example with Real Numbers

Take a cohort of eight customers whose first delivered order was placed in January. Measure repeat behaviour through March, giving every customer the same observation window. For this example, count delivered orders and exclude cancelled records consistently.

The customer-level data looks like this:

Repeat Purchase Rate Compared with Related Metrics

These measures answer different questions, so substituting one for another can lead to the wrong action. Customer repeat purchase rate asks whether customers came back at least once. The other measures describe how often they ordered, whether they remained active, or how much value they generated. Define the customer, order, and time windows before comparing results.

Repeat order rate is order-level rather than customer-level. A store with 100 orders, 25 of them placed by returning customers, could report a 25% repeat order rate under the definition “repeat orders ÷ all orders.” That does not mean 25% of customers returned: 10 customers could have generated those 25 repeat orders, while the remaining orders came from first-time customers. Because “repeat order rate” is used inconsistently, state its numerator and denominator in the report.

Purchase frequency measures order volume per customer, such as 180 orders ÷ 100 customers = 1.8 orders per customer during the period. It is useful when deciding how often to market, plan replenishment, or assess buying cadence, but it can hide the difference between many one-time buyers and a small group ordering repeatedly. Retention rate measures how many customers from a starting group are still active in a later period. It supports cohort health and churn decisions, but its answer depends on what counts as active and when the later period is measured.

Customer lifetime value (CLV) estimates or summarizes the revenue or profit associated with a customer over the relationship. It supports acquisition-budget and customer-profitability decisions, not just the question of whether a second purchase occurred. A customer with two low-margin orders may have a higher repeat purchase rate but lower CLV than a customer with one large, profitable order. For a fuller view, compare repeat behaviour with customer profitability analytics and the distinction between measurable and assumed inputs in Ecommerce Customer LTV.

The Mistakes That Make the Number Wrong

A changing denominator can make a stable customer base look better or worse. If 30 of 100 customers reorder, the rate is 30%. Add 50 new customers who have not had time to buy again, and the same 30 repeat purchasers become 20% of 150. The underlying repeat behaviour did not change; the reporting population did. Keep the cohort definition fixed, and record whether the denominator is customers acquired, customers with a delivered first order, or all customers active in the period.

Refunds and cancellations need an explicit treatment. A customer whose second order was placed and then fully refunded may count under a placed-order definition, but not under a realized-purchase definition. Mixing those rules overstates repeat behaviour and can also distort profitability. For a realized view, use delivered orders and inspect refund value alongside revenue; ecommerce return rate analysis explains why returned units and refunded value need separate treatment.

Subscriptions create another classification problem. A scheduled renewal is a second order, but it may not represent a customer choosing to buy again. Report subscription renewals separately when the question is voluntary repurchase. Include them when the question is whether the customer generated another completed order. The important point is not which rule you choose, but applying the same rule to every cohort.

Guest checkout and duplicate customer records can split one person into several customer IDs. For example, a first order under an account email and a second guest order under a different email can appear as two one-time customers, lowering the rate. The reverse problem occurs when shared household details merge different buyers, inflating it. Set an identity rule using the fields your store reliably captures, then review unusual records rather than silently changing the rule between reports. Shopify connections can sync orders, products and customers, but the identity definition still needs to be documented.

A fixed calendar window also introduces bias. A customer acquired on January 2 may have 90 days to reorder, while one acquired on March 30 has only two. Compare customers after the same elapsed time from their first delivered order, or label newer cohorts as immature. Do not mix January, February and March cohorts into one rate unless they have the same observation window and comparable acquisition mix. A blended figure can hide a genuine cohort change, such as a new channel bringing in customers with different repeat behaviour.

What to Do with the Answer

Use repeat purchase rate as a diagnostic, then segment it before deciding what to change. Keep the cohort definition and observation window fixed, and compare cohorts at the same age. Useful cuts include acquisition channel, first product or collection, first-order value, geography, discount use and customer type. A blended rate can hide the decision: customers acquired through one channel may reorder often, while customers who first bought a low-margin product may reorder but still destroy contribution.

Connect the rate to economics, not just customer count. For each segment, compare repeat behaviour with realized revenue, gross profit and contribution after advertising. A segment with a 40% repeat purchase rate is not automatically healthier than one at 25%: if its first and repeat orders each produce $4 after product cost and advertising, 40 repeat purchasers generate $320 of contribution across 80 customers. A 25% segment producing $18 per order can generate $900 across the same 80 customers when its repeat orders are included.

Timing determines payback. Suppose a first order contributes $12 after advertising and the customer costs $30 to acquire. The first order leaves $18 unrecovered. If a repeat order contributes $20, the customer reaches $32 cumulative contribution after the second order, so payback occurs on that repeat order. The repeat purchase rate alone cannot show whether the second order arrives soon enough, whether it is profitable, or how many orders are needed. Use customer profitability analytics to examine those customer-level economics rather than treating retention as a standalone target.

Monitor the rate by cohort age and segment alongside return value, product margin and inventory constraints. A high repeat rate may reflect a replenishment product, but returns can reduce the actual value of those orders; review how returns reach profit before increasing acquisition or retention spend. A low rate may indicate poor product fit, weak first-order economics or a purchase cycle that has not matured yet.

• If repeat rate and contribution are strong, test controlled scaling while checking payback timing and capacity.
• If repeat rate is strong but contribution is weak, fix product margin, discounting, shipping or advertising economics before pursuing more repeat orders.
• If repeat rate is weak but the first order is profitable, test post-purchase education, bundles or replenishment timing for the relevant cohort.
• If both are weak, investigate the product or acquisition source and stop treating the rate as a target to improve in isolation.

Repeat purchase rate is useful when its customer definition, cohort, and time window stay consistent. Calculate it from identified customers, validate the result against orders and margin, then use segments and related metrics to decide where retention work can improve profitability.