{"id":19854,"date":"2020-01-28T14:31:11","date_gmt":"2020-01-28T14:31:11","guid":{"rendered":"https:\/\/www.arimetrics.com\/glosario-digital\/cohort-analysis"},"modified":"2026-09-15T07:38:27","modified_gmt":"2026-09-15T07:38:27","slug":"cohort-analysis","status":"publish","type":"encyclopedia","link":"https:\/\/www.arimetrics.com\/en\/digital-glossary\/cohort-analysis","title":{"rendered":"Cohort Analysis"},"content":{"rendered":"

\"CohortDefinition:<\/strong><\/p>\n

Cohort analysis<\/strong> is a technique that groups users, customers, accounts or other units that share an event or characteristic during a defined period and compares how their behaviour develops over time.<\/p>\n

A cohort might consist, for example, of people who registered during the same week or made their first purchase in a particular month. The entry criterion and starting point must be defined<\/strong>, because they determine who belongs to the group and when its development begins to be measured.<\/p>\n

A cohort is not just any segment. In market segmentation<\/a>, membership may depend on current attributes and change over time; cohort analysis usually preserves a group formed around a shared event or period<\/strong> so that its trajectory can be compared.<\/p>\n\n

How cohort analysis is performed<\/h2>\n

The analysis begins with a question and a metric linked to it. The procedure can be summarised as five decisions:<\/strong><\/p>\n

    \n
  1. Define the objective:<\/strong> specify whether the analysis will study retention, repeat purchases, revenue, activation, feature use or another behaviour.<\/li>\n
  2. Set the entry condition:<\/strong> choose the event that adds each unit to the cohort, such as first registration, first purchase or the start of a subscription.<\/li>\n
  3. Choose the granularity:<\/strong> group by day, week, month or another interval suited to the cycle being observed.<\/li>\n
  4. Set the observation window:<\/strong> decide how many periods will be measured and which condition will count as a return or outcome.<\/li>\n
  5. Compare cohorts:<\/strong> display the values by elapsed time since entry and examine differences between comparable groups.<\/li>\n<\/ol>\n

    A cohort table commonly displays groups in rows and elapsed periods in columns. Each cell represents a cohort’s result at a comparable stage of its cycle<\/strong>, not necessarily on the same calendar date.<\/p>\n

    The definition of a user, event and period depends on the tool and the web analytics<\/a> implementation. Two platforms may produce different values even when they use similar labels.<\/p>\n

    Applications of cohort analysis<\/h2>\n

    The technique can be used for digital products, subscriptions, ecommerce and other services with repeated activity. These five applications are common:<\/strong><\/p>\n