Cohort analysis

A cohort is a group of individuals who share a defining characteristic or experience within a specified period. In cohort analysis, the cohort is typically defined by an event (e.g., birth, diagnosis, first purchase) and then followed over time. The method distinguishes between three effects: age (changes within a cohort as it ages), period (external influences affecting all cohorts at a given time), and cohort (differences inherent to the group). This framework is central to epidemiology and demography, where it helps disentangle these effects.

In business analytics, cohort analysis is used to measure customer retention, engagement, and lifetime value. By grouping users by acquisition date, companies can compare how different cohorts behave over subsequent periods, revealing whether improvements in product or marketing are effective. For example, a SaaS company might track monthly subscription cohorts to see if a new onboarding flow increases retention. Tools like Google Analytics and Mixpanel offer built-in cohort reports, making the method accessible to non-specialists.

Cohort analysis requires careful definition of the cohort and the time intervals to avoid bias. Common pitfalls include survivorship bias (if only active users are considered), selection bias in defining the cohort, and the confounding of cohort and period effects. Statistical techniques such as age-period-cohort models are used to separate these effects, but they rely on strong assumptions. In epidemiology, cohort studies can be prospective or retrospective, each with trade-offs in cost, time, and reliability.

Beyond business, cohort analysis has been applied in education to track student progress, in public health to study disease outbreaks, and in sociology to examine generational shifts. A notable historical example is the Framingham Heart Study, a long-running cohort study that identified major cardiovascular risk factors. In digital analytics, cohort analysis can be used to evaluate the impact of policy changes, such as GDPR, on user behavior. Also, cohort analysis is distinct from panel data analysis, which follows the same individuals over time, whereas cohorts may change composition.

Cohort

A group of individuals sharing a common characteristic or event within a defined period.

Retention

The proportion of a cohort that remains active over subsequent periods.

Age-period-cohort model

A statistical model that separates age, period, and cohort effects.