Still in beta, Google Analytics Cohort Analysis looks a little daunting at first. However, once you get your head round it you can gain some valuable insight into user behavior.
What’s a Cohort?
By using cohort analysis, Analytics presents a way of grouping users together according to their first website session. It works only according to the date the user was first acquired, and as such uses user tracking rather than session tracking.
The Cohort Analysis function groups together those who were acquired on the same date and takes a look at how their subsequent behaviour differed. It is possible to see how many returned to your site and eventually converted, and how many returned visits it took to do this.
Where to Find the Cohort Analysis Report
You’ll find it tucked away in the Audience section of your Analytics reporting window, right here:
When you first open the report, it will look something like this:
You have four cohort analysis options available, seen just above the graph.
Cohort Type: At the time of publishing, this option only allows ‘acquisition date’. This refers to the first time a user interacts with your site.
Cohort Size: This is where you can set different time frames for your cohorts; days, weeks or months. Again, at the moment this works on the first interaction model. For example, select weeks and cohorts will be grouped according to the first week they interacted, rather than first day.
Metric: When you first open up the report, this will be set to ‘user retention’. You can switch this to a list of other metrics by clicking the drop-down.
Date Range: This specifies the time range for which you wish to look at the cohorts. The option changes according to which cohort size you have chosen.
By default, the small button immediately above the graph will read ‘Pageviews per User’. You can switch this to compare up to four cohorts.
Reading and Using the Cohort Graph
The line chart shows the progress of cohorts according to your selected cohort size. By using that button next to the graph title, you can show the changes week on week for different acquisition dates.
Before you specific the cohorts for this graph, it will be set to the average of all of them.
The triangle chart you see below is a simple representation of users’ behaviour according to your selected metric. The darker the colour of each cell, the higher the metric value.
This example shows a date range of the previous nine weeks.
Quite simply, the x axis shows the average number of metric counts (in this case, pageviews) week by week following the initial interaction (or day or month, if you have chosen this as a measurement). The y axis shows each day, week or month by data at which the initial interaction took place.
For example, this triangle chart shows that returned pageviews were stronger in week 1 in the first couple of weeks of February than they were in the first few weeks of March.
Once you understand how to read it, it’s easy to use and can provide you with some pretty good insight when you line it up against the dates of your paid or organic campaigns.