Cohort analysis allows you to compare groups of users that are defined by shared characteristics. These can be things like: graduation date, geographic location, or performing a specific event. By defining these cohorts, you can analyze the their activity over time. For example: If you wanted to determine the average income for the graduating class of 2010, compared to all other graduates, you'd start by defining a cohort of users that graduated in 2010. You can read more about how to use Cohorts via the Mixpanel Support Documentation.

The challenge with this project was to create a tool that allows users to define a cohort that can be used within other Mixpanel reports. The final solution allows you to define which specific Events and Properties that would qualify users to be included in a cohort. You can even string together a series of events that must be performed in order (a funnel), for example: users that received a promotional email, logged in, and completed a purchase within 24 hours.

Cohorts can become very elaborate with the use of conditional statements: ‘or’, ‘and’, and ‘then’ – these allow you to be excessively granular with your definitions; for example: users from America (AND) users from Anywhere that have American citizenship (OR) users with no specified location that have a ‘@AOL.com’ email address.

For every large launch at Mixpanel the design team creates an enamel pin – as a way for the company to say 'thank you' – that represents the product or feature. For Cohorts I created a tessellating wolf pin; once connected to other wolf pins it creates a wolf pack (a cohort).