If you use Google Analytics you may have noticed a new report called Cohort Analysis.
This report helps you segment your data based on a date. Rather than looking at the individual user, Cohort Analysis looks at the patterns of groups of users that share common characteristics, attributes or experience.
How to use the Cohort Analysis Report
First, select a dimension in the Cohort Type. This is very straightforward as at present you only have one option, “Acquisition date”.
Next, select the Cohort Size. The cells show the number of users as a percent for the given time period. The darker the cell colour, the greater the percentage of cohorts. The top row shows the average for all cohorts relative to the time period. By selecting the dropdown next to “Acquisition date cohorts by user retention” you can compare up to four cohorts at a time.
Regardless of how large your date range is you will always see 12 columns. This is because the longer the time period the fewer insights you will get so there is little value of having more than 12 columns.
Cohort Analysis for Analysing Pay Per Click Device Traffic
By applying advanced segments to the Cohort Analysis report you can start to get some good data. Below shows a simple example of analysing customer retention for different device traffic.
We can extend this simple example by modifying existing system advanced segments to illustrate how different devices interact with AdWords pay per click traffic.
Copy the existing mobile and tablet advanced segments and add and additional condition where source / medium = google / cpc.
Copy the existing system’s “Mobile and Tablet Traffic” advanced segment. On the top filter change “Include” to “Exclude”. We have now excluded all mobile and tablet traffic which means it is only showing traffic from PCs.
When you apply your new advanced segments you should be able to see this:
From the results above we can see that by day five user retention for PCs has all but dried up. Tablet traffic has the highest retention (1.31%) followed by mobile traffic (0.26%).
What does this tell us? This suggests that people are comparison shopping using mobile and tablet devices and then using their PCs for the final conversion. You can see this behaviour when you look at transactions by device for the same time period. The table below shows us that the bulk of purchases come from PCs on the first day.
We can double check this finding by using the Multi-Channel Funnel reports.
In the report above mobile has an “Assisted / Last Click or Direct Conversions” score of 0.68 compared to desktops that have a score of 0.57. As the desktop figure is closer to zero this tells us that these devices convert more than they assist, which is what we would expect. Tablets are less of a “last converter” but there are still a sizable amount of last conversions from these devices.
Cohort Analysis Impact on Paid Search Strategy
So in terms of strategy how can this information help us? We would first recommend looking at costs. In the case of this website the mobile cost per conversion is far too high. Having now looked at the data we can see that this traffic is still important. It tells us that undecided customers are coming to the site via mobile.
What we need to do is to set a bid adjustment to reduce the costs by about -5%. We wouldn’t recommend reducing your spend further until you have about two weeks of data. If your costs are still high, reduce it by a further -5%. What you don’t want to do is to totally reduce mobile traffic as this can kill your overall conversions.