What is Cohort Analysis?

Cohort analysis is the study of users, customers, or employees within a defined cohort. The behavior of this group can help organizations predict the actions of similar cohorts in the future.

A cohort is any group of people described within an organization's data. To be considered a cohort, this group must:

• Have common attributes: All people in the cohort must have certain things in common, such as demographic details, location, or account type.
• Exist in a specific timeframe: Cohorts are bounded by time, so all members of the cohort data must exist between the specified start date and end date.
• Be described by data: There must be enough information in the data to perform relevant analyses for each cohort member.

Cohorts are similar to segments, except that cohorts are bound by timeframes. So, for example:

• "Website visitors from Ohio" describes a segment.
• "Website visitors from Ohio during Q1 of 2020" describes a cohort.

Other examples of cohorts include:

• New customer account signups during July
• Customers who spent over $10,000 in 2019
• Employees who worked overtime in January 2020
• People who submitted bug reports in Q1
• Visitors who arrived via organic search between 3/1 and 3/8

By focusing on cohorts rather than segments, analysts can account for situational factors, such as marketing campaigns, system upgrades, product launches, or economic conditions.

What is the Benefit of Cohort Analysis?

Cohort analysis is performed when analysts want to answer a specific question. For example, consider an organization that wants to find out if their marketing campaign has helped improve sales. They have data in their CRM that looks like this:

The analyst can define a cohort such as "all customers who ordered for the first time between 01/01/2018 ad 01/03/2018." Cohort analysis yields information such as:

• Average CLV (customer lifetime value) of this cohort
• Percentage of cohort acquired via marketing campaign
• Average CLV of cohort acquired via marketing campaign

The next step is to compare cohorts. The analyst might generate cohort data on first-time customers in 2019, for example, which allows them to discover data such as:

• Year-on-year growth
• Comparison of different marketing campaigns
• Changes in customer acquisition trends

This information can help to predict future user behavior. For example, the analyst may see that customers are more responsive to marketing campaigns during Q1, which is when department heads get their annual budget. The organization can then focus its marketing resources on making a big push during the most productive period.

Interpreting a Cohort Chart

A typical cohort chart will compare data from similar cohorts to identify trends that may not be otherwise obvious. For example, a company see consistent sales across the year, but they want to learn more about individual customer behavior. To do so, they perform cohort analysis on the customers acquired within each calendar month and see how many of those customers made purchases in the following months.

Their cohort chart might look like this:

This shows that the vast majority of customers make a purchase within the first month, but their business trails off over a six-month period. There's a notable drop after the third month, which might be the result of an identifiable factor, such as a three-month trial membership coming to an end. Using cohort analysis, this company can identify opportunities to improve retention and expand customer lifecycle.

Other Uses of Cohort Analysis

Cohort analysis can be useful in any situation where you want to analyze the behavior of small groups over defined periods. Common examples include:

• Analyzing the activity of users immediately after an upgrade to see if their behavior has changed.
• Tracking a real-life cohort, such as people who graduated together or new employees who started at the same time.
• Identifying customers last ordered in the previous year, to see if there is some common reason for losing them.

Cohort analysis is offered as a tool in platforms including Google Analytics and CleverTap.