Blog · Board and management reporting
An ecommerce repeat-customer benchmark needs the same purchase definition, follow-up period, and customer population as your report. Work through mature acquisition cohorts, a pooled rate, and a product-mix example before comparing an industry average.
An industry average repeat customer rate for ecommerce is useful only when its customer definition, purchase rules, and follow-up window match yours. Start with equally mature acquisition cohorts from your own order export, then compare a disclosed peer sample. An unexplained industry percentage is too weak a basis for a retention target.
A repeat-customer benchmark needs more than a percentage and an industry label. Ask who supplied the orders, which stores were included, how customers were identified, what counted as a repeat purchase, and how long each customer was observed. Check whether the result is a median store, an average of store percentages, or a rate pooled across all customers.
Even the ecommerce boundary can differ. The Census Bureau's retail ecommerce report defines ecommerce by how the order is placed or its terms negotiated; payment need not occur online. Its published estimates measure sales, rather than customer-level repeat purchases, so they cannot supply the customer-count benchmark calculated here.
For a retailer's comparison, also match the product category and buying arrangement. Replenishment items, replacement purchases, and recurring subscriptions do not give customers the same reason to reorder. A peer population should reflect those differences before its average becomes a target.
Use the guide to defining similar customers to decide which peers belong together. Industry is a starting boundary, not the complete definition of comparability.
For an acquisition-cohort comparison, use the first eligible purchase to assign a customer to a cohort. A repeat buyer places another distinct eligible order during the chosen follow-up period after that first purchase.
Cohort repeat customer rate = Customers with another eligible order during follow-up / Eligible first-purchase cohort customers × 100
This differs from a period report that counts customers with multiple orders inside the same calendar window. The general customer repeat rate formula covers that distinction and the order-counting method. Keep benchmark interpretation here focused on whether two rates describe the same population.
Set the follow-up period before reading the outcome. For the example below, it is the first 90 days after acquisition. A customer without another purchase by the cutoff has not repeated within that horizon; that does not establish that the customer will never return.
This is an observation problem. NIST's discussion of right-censored data explains that event times beyond a fixed test endpoint are missing. Applying that principle to purchases, a second order after your export ends is unobserved, rather than proof of permanent customer loss.
Use customer ID, original order ID, order date, order status, document type, and a supplied product category or acquisition channel. Resolve duplicate customer identities before assigning first-purchase dates. The first order visible in a short export is not necessarily the customer's first order with the business.
Remove duplicate order lines and return documents from the purchase-event list. Document whether fully refunded orders qualify, and apply that rule to both the first and repeat purchase. Separate automatically recurring orders if the peer benchmark covers only discretionary purchases.
Then group customers by first-purchase month and measure each customer's repeat event within the same follow-up horizon. Keep recent cohorts on a separate incomplete list until all customers have sufficient observation. Do not fix unequal exposure merely by putting an older and newer cohort in the same table.
A US online home-supplies retailer reviews orders through June 30, 2026. January, February, and March acquisition cohorts have each had enough time for the full 90-day follow-up. These are illustrative company results, not published industry averages.
| First-purchase cohort | Eligible customers | Repeat buyers within 90 days | No repeat within 90 days | Repeat rate |
|---|---|---|---|---|
| January | 200 | 60 | 140 | 30.0% |
| February | 300 | 75 | 225 | 25.0% |
| March | 500 | 100 | 400 | 20.0% |
| Total | 1,000 | 235 | 765 | 23.5% |
Each row reconciles: repeat buyers plus customers without a repeat equals eligible customers. The pooled rate is (60 + 75 + 100) / (200 + 300 + 500) × 100 = 23.5%.
The simple average of the displayed rates is (30 + 25 + 20) / 3 = 25.0%. That overweights January's smaller cohort. The customer-weighted check is 6.0 + 7.5 + 10.0 = 23.5 percentage points, using each cohort's customer share multiplied by its repeat rate. Percentages are shown to one decimal place; calculations use the underlying counts.
The declining sequence deserves investigation, but it does not establish worse service or a failed campaign. First check whether March brought a different product or acquisition mix. Also confirm that the export boundaries and customer identification rules stayed consistent.
Consider two product groups whose own repeat rates stay unchanged. The customer mix shifts between two equally mature periods.
| Period | Product group | Customers | Repeat buyers | Repeat rate |
|---|---|---|---|---|
| Earlier | Replenishment | 400 | 120 | 30.0% |
| Earlier | Durable products | 100 | 20 | 20.0% |
| Earlier total | 500 | 140 | 28.0% | |
| Later | Replenishment | 100 | 30 | 30.0% |
| Later | Durable products | 400 | 80 | 20.0% |
| Later total | 500 | 110 | 22.0% |
The overall rate falls by 28.0 − 22.0 = 6.0 percentage points even though neither product group's repeat rate changes. The later period contains more durable-product customers. Comparing either total with an outside average without its product mix can produce the wrong diagnosis.
Ecommerce operations should own the order eligibility rules and customer identity checks. Marketing should supply acquisition cohorts and channels. Finance should confirm that the chosen repeat behavior supports contribution after discounts, returns, and service costs.
A commercial owner can then set a target for comparable groups rather than chasing the company average alone. Use dormancy by industry to distinguish normal purchase gaps from unusual inactivity. Keep the question of customer lifespan by industry separate: observed repeat buying over a fixed horizon does not estimate a customer's remaining lifetime.
Bring an order export with identified customers, dated orders, statuses, and supplied cohort or product fields to Covirage. Covirage maps the columns, and its tools compute and check the repeat-buyer counts and eligible populations; the external AI model explains the comparison with every figure traced to its rows. Create a free Covirage account and bring a sample of your own export.
See renewal watch for the supporting customer analysis.
Use a target based on comparable mature cohorts and the product's expected reorder cycle. An outside average needs a disclosed sample, observation window, and repeat definition. Without those details, the difference between your rate and the published percentage is difficult to interpret.
Sources may count different events, customers, stores, and follow-up periods. Some measure customers with multiple orders during a calendar period; others measure new customers who reorder after acquisition. Subscription-heavy and occasional-purchase retailers can also have very different buying patterns.
They belong in an acquisition cohort, but compare that cohort only after the chosen follow-up period has elapsed for every included customer. Otherwise recent acquisitions have less opportunity to reorder, and the benchmark gap partly measures incomplete observation.
Usually not. Cohorts of different sizes require a customer-weighted calculation, and calendar-month populations can overlap. For nonoverlapping acquisition cohorts with equal follow-up, add repeat buyers and eligible customers separately, then divide the totals.