Cohort analysis groups customers by a shared starting point, usually the month they were acquired, and tracks how each group behaves over time. It reveals whether retention, spend, and value are improving or declining for newer customers versus older ones, which a single blended average hides.

Cohort analysis is how you tell real improvement from a flattering average. Blended numbers can look fine while every new cohort retains worse than the last, which means the business is quietly decaying. By tracking each acquisition group separately, you see whether the changes you are making actually move retention and value, or just get masked by your older customers.

Example:

 Customers acquired in January retain 90% at month three; those acquired in April retain only 75%. The blended number looks stable, but the cohort view shows newer customers are leaving faster.

What does cohort analysis reveal that averages hide?

 Whether newer customers behave differently from older ones. A stable blended average can mask declining retention or value in recent cohorts, which signals a worsening business.

What is a common way to group cohorts?

 By acquisition month, so you can compare how each month's customers retain and spend over their lifetime. Cohorts can also be grouped by channel, plan, or campaign.