A store can report improving advertising ROAS while its new-customer acquisition becomes more expensive. Repeat buyers may account for a larger share of attributed orders, especially during replenishment campaigns or promotions sent to an established customer base.
Repeat revenue is valuable. The measurement problem is allowing it to answer a different question: how effectively the business is acquiring new customers. Separate acquisition, retention, and attribution so each can inform the decision it actually supports.
Establish the customer and first-order definition
Choose the customer identifier and define which order qualifies as the first purchase. Decide how to handle canceled orders, test orders, duplicate profiles, guest checkout, and historical imports. Record limitations rather than assuming identity is perfect across systems.
The first-order date should anchor the acquisition cohort. Each subsequent eligible order can then be classified by its age relative to that first purchase. Keep the underlying order ID so refunds and adjustments can be reconciled later.
Shopify's customer reports documentation describes first-order cohorts and notes that some customer reports use a customer's broader order history. Inspect the chosen report's definition; a customer who returns later can affect classifications differently from an order-level first-versus-repeat field.
Separate period reporting from cohort reporting
A period report asks what happened this month. It may include first purchases from new customers and repeat orders from many older cohorts. A cohort report asks what happened to customers acquired in a particular period as time passed.
Both are useful. Period reporting helps manage current revenue and operations. Cohort reporting helps assess acquisition quality and payback. They should not be combined into a single unexplained lifetime-value number.
Use a simple structure:
| View | Useful question |
|---|---|
| New customers acquired this month | What did current acquisition produce? |
| Repeat orders placed this month | How much current demand came from existing customers? |
| Month-one contribution by acquisition cohort | Are similarly mature customers developing differently? |
| Cumulative contribution by cohort age | How quickly is acquisition spend being recovered? |
Compare cohorts at the same age. A six-month-old cohort has had more opportunity to repurchase than one acquired last week.
Keep attributed orders distinct from acquired customers
An advertising platform may attribute a repeat order to a recent click or view under its reporting rules. That does not make the purchaser a newly acquired customer, and it does not establish that the order would not have occurred otherwise.
Report the platform-attributed result alongside the store's customer classification. Where the systems cannot be joined reliably, show the gap rather than allocating every unidentified order to new acquisition.
The attribution-versus-incrementality guide explains why these questions differ. Use an appropriate experiment when the decision requires a causal estimate of retention advertising, rather than assuming every attributed repeat order is additional demand.
Examine a mixed-customer example
Suppose an illustrative store spends $10,000 in each of two months. It records 200 new customers in the first month and 160 in the second, while repeat orders increase enough to keep total attributed revenue stable.
Under a simple spend-per-new-customer calculation, acquisition cost rises from $50 to $62.50. A stable blended ROAS would not reveal that change. The example does not prove the ads became worse; channel allocation, tracking, demand, and customer mix still need investigation.
Use the new-versus-blended CAC worksheet to define the spend scope. If some spend is explicitly devoted to retention, report that allocation rather than pretending all advertising dollars served only acquisition.
Measure repeat contribution after its costs
Repeat order revenue may include a discount, shipping subsidy, loyalty benefit, or a lower-margin product mix. Calculate contribution using the agreed cost model and allow for returns where the data support it.
Track the marketing costs used to generate repeat demand as well. A customer is not costless to serve or re-engage just because the first acquisition expense occurred earlier.
Avoid projecting a long lifetime from a short replenishment pattern. Show observed contribution separately from modeled future value, with the assumptions that drive the model. The cohort payback guide offers a related framework for recurring businesses.
Inspect why repeat behavior changes
Look at product replenishment cycles, promotion timing, stock availability, fulfillment quality, and customer experience. A spike in repeat orders may be a temporary pull-forward from next month's demand rather than a durable retention improvement.
Segment only where the data can support a useful comparison. Product category, acquisition offer, and customer region may reveal meaningful differences, but dozens of sparse segments can produce unstable stories.
Keep customer communication permissions and privacy requirements within the business's established process. A measurement classification is not authorization to contact every person in the cohort.
Build a decision-ready reporting cadence
Show current new-customer volume and acquisition cost, current repeat contribution, and a small set of equally mature cohorts. Include the latest data cutoff and unresolved identity or refund limitations.
When blended results improve, ask whether the improvement comes from acquisition, retention, order economics, or a reporting change. When acquisition weakens, use the cohort view to determine whether later customer value supports the cost or merely reflects optimism.
This structure lets the business value repeat customers fully while keeping acquisition performance visible. It also makes advertising decisions easier to explain without asking one blended metric to represent every stage of the customer relationship.