Measurement Planning
Part of Customer cohort analysis for acquisition-labelled ecommerce value
Grouping customers by first completed purchase
Define a completed purchase, resolve customer identities and assign first-purchase cohorts without confusing first-order date with payment confirmation.
Define the qualifying milestone before assigning a first completed-purchase cohort. Find each resolved customer's earliest order that meets it, then assign that order's date to a week or month. Preserve the order and identity evidence behind the assignment. A platform's first-order field may use a different milestone.
Fix the entry rule
Choose what completed means: payment confirmation is one possible rule; an order merely placed is another event. Under payment confirmation, an unpaid order does not start the clock. Decide whether a later cancellation or refund changes the original assignment or only the value reported for that cohort. Apply one documented rule to every customer.
Shopify's Customers reports include a Customer cohort analysis section. Its New vs returning customer report defines a first-time customer as someone who placed their first order. Shopify sales reports include pending, cancelled and unpaid orders. For a custom payment-confirmed rule, record the qualifying status, timestamp field, store time zone, test-order exclusions and treatment of later changes before assigning cohorts.
Defining and Assigning First Completed-Purchase Cohorts
- Define 'completed purchase'Choose a clear rule (e.g., payment confirmation) to determine when a purchase is complete.
- Record the qualifying timestampCapture the date and time of the first valid completed purchase, including store time zone.
- Apply consistent rulesUse one documented rule across all customers—do not allow exceptions without justification.
- Handle later changesDecide whether cancellations or refunds affect the original cohort assignment or just reporting value.
First Purchase vs Payment Confirmation: Key Differences
- Event Trigger
- Order placed (even if unpaid)
- True Completion
- Payment confirmed by system
- Impact on Cohort Date
- May differ significantly if order is abandoned or refunded after placement
- Reporting Accuracy
- Payment confirmation gives more accurate customer acquisition timing
Resolve the customer key
Use a documented identity rule to link orders. Keep the platform customer ID and any approved links used to combine account or guest records. Shopify's customer guidance says a profile can be created when someone places an order or starts an order but abandons checkout, so a profile alone is not evidence of a qualifying completed purchase.
Place orders that cannot be linked confidently in an unclassified group. An unmatched guest order is not proof of a new customer. Report how many orders or customers could not be classified, because missing identity can change both cohort size and repeat counts.
Assign and check the date
For each resolved customer, sort qualifying orders by the chosen milestone timestamp. Use a stable order-ID tie-break if timestamps match. The earliest qualifying order is the entry order; a later qualifying order is a repeat, even if both fall in the same calendar month.
Store the customer key, entry order ID, timestamp, cohort period, rule version and extraction date. Check records representing an unpaid order followed by a paid one, a later cancellation, two orders in one month, a linked profile and an uncertain guest order. The expected classification depends on the chosen rule and actual records.
If a late update changes which order qualifies, recalculate the affected assignment under a recorded rule version and explain the revision. This assignment record supports a cohort comparison without turning today's customer profile or report label into an unexamined historical fact.
