Short answer
Group customers by first-order period and compare repeat buying at the same age. Separate immature cohorts, seasonal cycles and revenue from contribution.
The answer in detail
A customer cohort groups people by the period of their first order and follows their later purchases. Shopify already offers a cohort report. Compare January's customers at month three with February's customers at month three, rather than comparing both groups in the same calendar month.
The benchmark
An overall repeat-purchase rate mixes customers acquired two years ago with people acquired last week. It cannot tell you whether recent customers return more reliably than earlier ones.
The Shopify documentation cited in the source describes rows grouped by first-order week, month or quarter, with columns showing elapsed time. Available measures include retention, sales and spend per customer; a chart can overlay cohort curves.
The source contrasts this with GA4's cohort exploration, which Google describes as device-based. One person's desktop purchase and later phone purchase may not be linked in that exploration. Shopify customer profiles answer a different identity question, although duplicate profiles can still need attention. Do not expect the two systems to produce identical counts.
This fictional table shows the share of the original cohort purchasing in each month, not cumulative repeat purchase:
| First-order month | Customers | Month 1 | Month 2 | Month 3 | Month 6 |
|---|---|---|---|---|---|
| January | 820 | 6% | 4% | 4% | 3% |
| February | 760 | 7% | 5% | 4% | Not yet mature |
| March | 1,140 | 4% | 3% | 2% | Not yet mature |
In an anonymized lingerie-and-basics case reported by the contributors, value per customer ranged from €170–€230 in the first-order month and €250–€300 at 12 months. Full-price cohorts outperformed sale cohorts. Those are specific reported account results, not reference targets.
The rule
Align cohorts by age, then interpret the curve in the context of the category. January's third month and February's third month occur at different calendar dates but represent equal time to return.
A curve often falls after the first period and then stabilizes. A persistent repeat-buying level can indicate a recurring customer base, while continuing decline calls for investigation into product fit and experience. Neither shape proves a cause by itself.
Seasonal products need an appropriate clock. A winter-accessories brand in the source recruited mostly in December, often for gifts, and only a few hundred customers in summer. Its winter 2024–25 cohort had 12.5% repeat buyers, with about 5% returning the following winter. A short monthly view could mistake normal seasonality for permanent loss. Allow a full buying cycle before judging it.
The steps
- Open Shopify Analytics > Reports and find Customer cohort analysis. Confirm the report and access available in your account rather than relying on a translated menu label.
- Start with monthly cohorts and retention. Use quarterly intervals if purchases are infrequent, and inspect both the table and the overlaid curves.
- Reconstruct the table when you need a custom definition. With customer ID in column A and order date in C, English Excel uses
=MINIFS(C:C,A:A,A2)for first-order date in D2. Calculate calendar-month age with=(YEAR(C2)-YEAR(D2))*12+MONTH(C2)-MONTH(D2)in E2, and cohort with=TEXT(D2,"yyyy-mm")in F2. Deduplicate customer-and-age pairs, pivot cohort against age, and divide unique repeat buyers by the original cohort size. Formula separators depend on spreadsheet locale. - Read rows and calendar effects separately. A diagonal pattern each November can reveal a promotion bringing back multiple cohorts, including holiday buyers.
- Segment by acquisition channel, first product, destination market and discount. In a drop-based apparel case, 69% of 1,941 newly acquired customers had not returned, and 78% of second orders occurred outside drop dates. Check the observation window before drawing a conclusion.
- Review new cohorts monthly. Compare month-one behavior with earlier cohorts when assessing a second-purchase sequence, then revisit mature periods.
The pitfall
Period 0 is not simply the first order. Shopify's cited definition includes repeat orders within the same initial period. Someone buying twice in January can already appear there. Use subsequent periods to understand later retention.
An unavailable period is not zero. September has no month-six result in October. Mark immature cells clearly, as in the repeat-purchase-rate method.
Revenue also flatters value. Convert each cohort to contribution before comparing acquisition cost. In the lingerie case, ratios approaching ten reflected incomplete acquisition costs. Shopify's projected values with sufficient history remain forecasts, not guaranteed sales or a validated LTV:CAC ratio.
What to tell your team
“Our fictional March promotion recruited 44% more customers than the January–February average, but only 2% repurchased in month three, versus 4% for those cohorts. The next test will feature an entry product associated with later repeat buying, and we will compare contribution at equal age.”
