Advanced Mobile Cohort Analysis using Adobe Report Builder
In the previous post Cohort Analysis was introduced and we set up Classifications to be able to do analyze App Install Cohorts in SiteCatalyst. This post will continue from that point and focus on using Adobe Report Builder for robust Mobile Cohort Analysis.
If your App Install date eVar expiration has not been set to “never” use the following; otherwise, skip to section B.
Section A, using Segments for Cohort App Install Month Analysis
If the eVar capturing App Install Month is not set to expire “never”, then you will need to use SiteCatalyst segments to get the data for each cohort. Create a Visitor segment for each monthly cohort where “App Install Month equals (some month)”. Each segment should look like the example below.
Then create a data block in Excel using the Report Builder tool and the LifeCycle metrics available from the AppMeasurement libraries as well as any engagement metrics used for the app. Be sure to use the visitor segment for the App Install Month and create a data block for each month’s cohort.
Use Excel formulas to reorganize the data to appear like the data blocks below that highlight one metric per data block, and where each cohort is on a different row.
Section B, using App Install eVar with “never” expire for Cohort App Install Month Analysis
In Report Builder create a data block of 12 Month Cohorts trended over 12 months. In the example below App Install cohorts from 2012 are trended across all of 2012 by month. The data block starts in cell D10 with cohorts running down column D, while month data is trended across columns E through P. Use the LifeCycle metrics from the AppMeasurement library or any event you are tracking in the app.
Arrange the data using Excel formulas so that the columns contain metrics for “1 month after install”, “2 months after install”, etc. by copying the Cohort names from D27 through D38. And label “Month 1″, “Month 2″, etc. across row 27. Next, reorganize your data to pull in the metric according to the length of time it occurred after Install. If you have followed the Excel locations used in the examples above, you may use the formula: =IFERROR(INDEX(E12:P23,ROW($D27)-ROW(D26),ROW($D27)-ROW(D26)+COLUMN(E$26)-COLUMN(E26)),”-”). This formula can be copied/pasted or dragged across from E27:P38 and will dynamically populate all the metric data for your cohorts across the entire time period.
Add Excel’s Conditional Formatting Color Scales to the chart, along with any other formatting changes you prefer and you have Cohort data you can start to analyze! The metric for App Launches was used in the examples above, but any event can be used in Cohort Analysis and may provide different clues for your Mobile App’s user experience.
For example, when you create cohort data for launches per month, you can start to see how many months after download the user engagement with app begins to decline. In the example you can see that around 6–8 months after install users start to launch the app less often. Consequently, the 6–8 month time period could signify an opportune time to start to re-market app features or upgrades to users to keep them engaged. It could also signify the life span of the app and allow marketers to adjust spend accordingly depending on the app itself.
Looking at churn rate by Cohort allows marketers to keep an eye on app loyalty by segment. From the data we can see that our January cohort (which in the example is the first group of App Installers) does not reach a churn rate of 65–70% until 10–12 months out. While our July and August cohorts are reaching that same churn rate at only 4–5 months out. This is showing that our early adopters are not only installing the app earlier but also sticking around longer. You may be familiar with the innovation adoption curve with “Early Adopters” engaging readily, followed by “Majority Conservative” and finally “Skeptics”. Cohort analysis may help you make decisions on whether this applies to your app and how to market to the different cohorts.
Cohort Analysis was also a highlight of the Advanced Mobile Analytics session at Summit 2013, you can click through to view the session and more!