Measurement Planning
Part of Ecommerce customer value analysis
Estimating customer value from observed purchase history
Reconstruct observed customer value from qualifying orders, a fixed time window, later adjustments and the original entrant count.
To calculate observed customer value from order records, identify each eligible entrant, attach qualifying orders within a fixed elapsed window, apply adjustments once and divide by the original entrant count. The result describes recorded value through that window, not future purchases.
Build one record per entrant
Choose the qualifying first-order milestone, such as order placement or payment confirmation, and a documented customer identity rule. For each resolved customer, keep the entry order ID and timestamp.
Exclude test orders under a stated rule. Keep orders with uncertain identity in an exception total; an unmatched order is not evidence of a new customer.
Define the time boundary precisely. For a 90-day window, one workable rule includes orders from the entry timestamp up to, but not including, 90 elapsed days later. Use the same time zone and rule for every entrant.
Include a customer in a completed-window result only after that full period has elapsed.
Steps to Estimate Observed Customer Value
- Define first-order milestone (e.g., payment confirmation)
- Apply customer identity rule and exclude test orders
- Set fixed time window (e.g., 90 days from entry timestamp)
- Include only completed windows in final count
- Reconstruct value with adjustments (refunds, reversals)
Reconstruct the amount
Write the value formula before adding orders. Product sales after recorded discounts and relevant product-value reversals differ from a total that includes tax and delivery. Contribution also needs cost records; missing costs remain unknown.
Keep each qualifying order ID once. Associate any later adjustment with its original order, even if the adjustment was processed after the 90-day purchase window. State the adjustment cut-off. If an exported amount already reflects a refund, do not subtract the refund again.
Consider two hypothetical eligible entrants. The first has an A$100 entry order and an A$40 later order inside the window; A$20 of that later order is refunded by the stated cut-off.
The second has an A$50 entry order and no later order. Their combined adjusted value is A$170, so observed value is A$85 per entrant. The denominator is two, not the one customer who returned.
Shopify's sales reports list gross sales, discounts and sales reversals as separate terms, and define gross sales using product price and quantity before taxes, shipping, discounts and sales reversals. Pending, cancelled and unpaid orders are included in gross sales. Select orders against the population required by your calculation rather than relying on a report label.
Publish the calculation with its coverage
Show eligible entrants, qualifying order count, total adjusted value, value per entrant, currency, extraction date and adjustment cut-off. Show unmatched order value separately. A median or distribution can reveal when a few customers dominate the average.
Shopify's customer-report list includes Customer cohort analysis, and Shopify says customer reports can provide average order count, average order totals and expected purchase value. Check the report's definitions before comparing it with a custom calculation. Keep any future projection separately labelled.
