Customer Cohort
A customer cohort is a defined group of customers sharing a starting event or characteristic that can be followed consistently over time.
What is Customer Cohort?
In ecommerce, a common cohort consists of customers whose first purchase occurred in the same month. The group is established at that starting point and then followed for repeat purchases, contribution, returns or acquisition recovery.
A cohort differs from a changing segment such as “customers active this month.” The segment’s membership can change whenever behavior changes. A first-purchase cohort keeps its original membership so later performance can be compared against a stable starting group.
Following January buyers into later months
A store acquires 200 first-time customers in January. Fifty of them buy in their second month after acquisition, giving a month-two purchase incidence of 25%. New customers acquired in February are not added to January’s cohort, even if they buy similar products.
How to interpret it
A cohort provides a comparison structure rather than one standalone performance formula. Different metrics can be calculated for the same group: revenue per original customer, repeat purchase rate, return rate or cumulative contribution. Each still needs its own definition.
Compare cohorts at equivalent ages. January customers have had more time to order than customers first acquired in June. Comparing their cumulative revenue on the same calendar date gives the older group an automatic advantage.
Decide whether a later interval means a specific period or any activity up to that point. “Bought in month two” is not the same as “bought again by the end of month two.” Both are useful, but they answer different questions.
Cohorts can also be defined by acquisition channel or first product, ideally using stable rules recorded at entry. Overlapping groups are possible, so totals should not be added as if every cohort were mutually exclusive.
Keep the original denominator
A customer who never orders again remains part of the acquisition cohort. Removing inactive members would make later averages describe survivors rather than the customers the business originally paid to acquire. The stable denominator is what lets cohort analysis reveal weak repeat behavior instead of hiding it.
Common mistakes
Comparing calendar totals instead of customer age
A mature cohort and a recent cohort need equal observation windows before their repeat or cumulative value metrics are compared.
Changing membership after seeing the outcome
Excluding refunds or one-time buyers after the fact can bias the result. Define eligibility before interpreting performance.