The Basics Of Cohort Analysis

If you hadn’t heard of cohort analysis before, you probably have now that Google is offering a beta version as part of their analytics suite, but do you know what it could mean to your business?

What Is A Cohort?

Firstly, let’s start with a simple definition; A cohort is a group of people who share a common characteristic within a defined period of time. For example, all users who first visited your website in March 2014.

Cohorts are used a lot in medical studies, so a group of people that have been exposed to a certain drug or pollutant can be put into a cohort based on when they were first exposed. Or a group of people all born in 1984, could form a birth-year cohort.

A cohort is simply just another way to segment a group of people in order to analyse behaviours.

Cohorts In eCommerce

Here is a general example for us to work with; An eCommerce business might be interested in how customers who first purchased through them on Black Friday in 2014 behave compared to other customers. This would become a ‘Black Friday 2014’ cohort.

The common characteristic is that they first made a purchase and the defined period of time is Black Friday 2014.

The business could then monitor the behaviour of this group over a longer period of time to learn more about their buying habits, the long-term value of that group and key differences when compared to other cohorts.

Let’s consider another example; An eCommerce business that provides ski equipment will have very clear seasonality trends. There are some easy ways to use cohort analysis in order to identify more long-term patterns for certain cohorts of customers.

By setting up ‘End of Season/Year’ and ‘Beginning of Season/Year’ cohorts, the business could monitor the behaviour of those customers over a long period of time in order to learn how best to communicate with them, market to them and retain them.

Cohort Analysis Using Google

Now for the bad news. Cohort analysis is currently difficult using Google Analytics.

The main reason for this is that we need the date of something impactful like a conversion to be able to create our cohorts and currently Google doesn’t give us an activity date for reports or segmentation. The only available dimension currently is the acquisition date.

Another reason is that Google segmentation is based on visits and not on users so even if we did manage to set up a date-specific, action-based cohort (perhaps using UTM codes), we would only be able to see visits that matched our criteria rather than all the visits of the users that matched our criteria.

It’s worth noting that cohort analysis and user segmentation is the direction in which Google analytics,m and analytics in general, is heading. Google Remarketing is incredibly popular and completely reliant on segmenting a site’s audience based on behaviour…cohort analysis works in a very similar way.