Cohort Analysis in Google Analytics

Have you begun using the new Cohort Analysis available in Google Analytics?  What are you waiting for?  There is a wealth of valuable information there about the behaviors of the most important people in your virtual world: the folks who are actually visiting your website!  It’s really a fairly simple report for one the can give you so much actionable information.  Let’s take a look at how Google Analytics has set this up, what the reports can analyze, and how you can use that information to improve your overall online marketing efforts.

Defining The Term

The word “cohort” comes to us from the ancient Roman Army, where it was the most basic military unit.  A Cohort was made up of 480 soldiers; ten Cohorts comprised a Legion.  In the modern Google world, “cohort” has been adapted to refer to a group of people sharing a similar trait, whose behavior on a website is being tracked as a group over time. Presently, while Cohort Analysis is still in the beta stage, the only shared trait that you can consider is the date of a visitor’s first session on a website.

You could look at groups whose first session occurred in the same month, in the same week, or on the same day, so an individual is theoretically a member of three cohorts.

How Does It Work?

You’ll find the Cohort Analysis option under the Audience section of Google Analytics.  When you open this report, you’ll be met by a two-section screen.  The top section gives you some options to clarify the information you want to see, and a linear graph:

5.29.15 - Cohort (Image 1)

Choose your cohort type (again, at present you may only choose to group by acquisition date) and size (day, week, month).

The metric drop down gives you options of the data you want to track for each cohort over time.  If you are running an EComm site, you may be most interested in revenue generated by each cohort, or number of transactions; many website owners find the default measurement, “user retention,” to be valuable – this measures the percent of users from each cohort who return to the site in succeeding days.  This is the metric we’ll use in this example.

Finally, you’ll choose to measure that data over a time span of 7, 14, 21 or 30 days.

The linear chart graphs the average data of all cohorts by default; however, by using the drop down at the top of the graph, you are able to graph one cohort in comparison to others.  Still, the linear graph can be difficult to interpret.  That’s where the second part of the screen comes in, presenting the data in a chart that compares each cohort in a sort of “heat map” where better results are shaded more heavily:

5.29.15 - Cohort (Image 2)

You can quickly see which cohorts had better short term and long term results.  In our example, you can see that the cohorts of May 16, May 19, May 24 and May 27 had the strongest retention on Day 1 (the first day after their acquisition date). It begs the question, what was happening on those days that caused more people to come back to the site the next day than any other?  Do those dates coincide with an email blast?  New content posting?  An on-site contest or game of some sort?  Whatever it may have been, if a pattern is found it may be a pattern you want to consider using even more often.

Go Beneath the Surface

But let’s look further.  In all four of those cases, the drop in retention to Day 2 is precipitous.  The cohorts of May 17 and May 25 didn’t have as strong an immediate retention, but a fair number of those folks came back the next day – and the day after that as well. And the cohort of May 15, despite dipping a bit on Day 2, was still showing string retention through Day 4.  You would really want to look back at those dates to see what factors may have contributed to those successes.  If you wish, you can dive even deeper into this data by segmenting out specific types of traffic.

It’s worth spending some time familiarizing yourself with Cohort Analysis.  In its beta form, it already poses the kinds of questions you want to be asking yourself about what is and isn’t working on your website. Undoubtedly as Google rolls it out from beta, additional traits for defining cohorts will be added and a wealth of information will be unlocked.

If you are looking to partner with a team of experts to help analyze and optimize your website, start a conversation with Synapse today!


This entry was posted in Marketing, News. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *