Updated: Jul 23, 2026 • 3 min read
Automate ecommerce customer cohort reporting
A blended repeat purchase rate can hide whether newer customers are becoming more or less valuable. Cohort reporting groups customers by their first purchase period, then tracks how each group returns and spends over time.
Why cohort reporting is more useful than a blended average
If a brand acquired a large number of customers during a promotion, its total returning customer rate can change even when customer behavior did not. A cohort view separates acquisition mix from retention performance.
Shopify's customer cohort analysis documentation describes cohort views for retention, repeat purchases, sales, average order value, amount spent per customer, acquisition channel, product, and subscription status.
The recurring operational problem is not access to a cohort table. It is turning the table into a consistent explanation of:
- Which cohorts improved or deteriorated
- When customers normally place the second order
- Which channels produce durable customers
- Whether discounts acquired low-retention buyers
- Which segment should receive a retention experiment
Data to connect
Use read-only access to:
- Shopify customers, orders, refunds, and first order dates
- Acquisition source and campaign data
- Klaviyo or another lifecycle platform
- Subscription data, if applicable
- Product, collection, and discount attributes
Agree on customer identity rules before calculating cohorts. Guest checkout, changed email addresses, and merged profiles can otherwise split one customer into several records.
Metrics and comparisons
For each monthly or weekly acquisition cohort, calculate:
- Customers acquired
- Second purchase rate
- Retention rate by period
- Orders per customer
- Average order value
- Net revenue per customer
- Time to second purchase
- Refund-adjusted customer value
Break out results by first order channel, campaign, product, discount, and subscription status only when the cohort is large enough to support a useful comparison.
Example Agent instruction
"On the first business day of each month, update the customer cohort report. Group customers by first purchase month and show customer count, second purchase rate, repeat revenue, orders per customer, average order value, refund-adjusted value, and median time to second purchase. Compare mature cohorts at the same age. Identify the three largest changes and link each conclusion to its source data. Do not compare a three-month-old cohort with a twelve-month-old cohort as if they had equal observation time."
What the Agent should produce
A useful output includes:
- A cohort grid with like-for-like age comparisons
- A short retention narrative
- The strongest and weakest acquisition cohorts
- Channel and first-product observations
- A list of segments ready for lifecycle testing
- Missing identity or attribution data
The Agent can prepare a Klaviyo segment or campaign brief, but a marketer should approve any customer-facing message.
Implementation workflow
- Define customer identity, first purchase date, net revenue, and refund treatment.
- Connect Shopify and lifecycle sources through Connectors.
- Reproduce two historical cohorts and compare with Shopify.
- Add campaign and first-product dimensions after the base totals reconcile.
- Schedule the monthly report in a Document.
- Route material changes to retention and growth owners.
- Review Logs when cohort membership changes unexpectedly.
Review checklist
- Cohorts use the same definition every period.
- Comparisons use equal cohort age.
- Refunds and canceled orders follow the agreed rule.
- Customer identity issues are visible.
- Small cohorts are not presented as strong evidence.
- Recommendations name an owner and a testable next step.
How to measure success
Track report preparation time, identity match rate, time from insight to experiment, second purchase rate, and the percentage of recommendations that become measured retention tests.
Next step
Begin with monthly cohorts and one retention metric. Once the numbers reconcile, add acquisition channel and first product so the report can guide specific lifecycle experiments. Book a demo to design the first version.