Instead of looking at all customers as one blended pool, cohort analysis compares groups that started at the same time or in the same way. A typical eCommerce cohort table lists first-purchase months down the side and months since first purchase across the top, with each cell showing repeat purchase rate or cumulative revenue per customer.
You can also build cohorts by acquisition channel, first product bought, discount used on the first order or campaign. This answers questions blended reports cannot: Do customers acquired through Meta repurchase as often as those from Google Search? Do customers who first bought on a big promotion stick around?
Why it matters for eCommerce
Cohorts show whether retention is improving or quietly getting worse, long before it appears in total revenue. They also give you a realistic customer lifetime value by channel, which tells you how much you can afford to spend on acquisition and still make a profit.
How it works in practice
- Export orders with customer ID, order date and value from Shopify or your data warehouse.
- Group customers by first-order month, then calculate repeat rate and revenue at 30, 60, 90 and 180 days.
- Split cohorts by channel, first product or first-order discount to find your best customers.
- Use the results to set CAC targets and to prioritize retention flows.
Common mistakes
- Judging recent cohorts before they have had time to repurchase.
- Using revenue instead of contribution margin when estimating cohort value.
- Mixing guest checkouts with duplicate customer records, which understates repeat behavior.