You launched your promising mobile app a few months ago and now find yourself dealing with retention, monetization and ongoing efforts to increase user stickiness. There is plenty of data available to you: how many installs took place, how many users actually opened your app and unfortunately, also how many uninstalled it. You can even segment your users demographically. This is all important information, but it is far from enough.
Let’s take a running app as an example. We’ll name it RunApp and examine two users: Mark, a fifteen-year-old man from New York City and Ana, a fifty-year-old woman from Madrid. These two users differ in almost every demographic parameter, but they behave the same way – they take the same actions. Both of them run more than three times a week and use the app frequently. They are in the same cohort – users who run more than three times a week. What can we learn from this? How will it help us to improve RunApp’s performance? We’ll get back to Ana and Mark later in this post, but first let’s explain a few things.
To get the most out of your data and to best analyze it, you must segment your users in a manner which would enable you to draw the most useful conclusions. We call this behavioral segmentation. We also recommend a commitment to an endless cycle of measuring, learning, applying changes and then measuring again. Analyze your data using Cohort Analysis. You’ve probably heard about it before, but what is it, really?
A cohort is a group of people sharing a common characteristic, like activity, feature, or behavior, over a set period of time. Cohort analysis addresses the why and how of user activity – why users performed certain actions (open, click, buy, etc.) and how they did it, namely, what was the sequence of events that led to any particular action. Unlike the traditional way of looking at raw data, cohort analysis provides a deep dive into groups of user behaviors. Opening the app, making an in-app purchase, upgrading from free to premium, abandoning the app, uninstalling… these are just a few examples of analyzed behavior.
In the mobile app ecosystem, cohort analysis presupposes that users’ behavior changes over time and provides tools to assess the lifecycle of a user within the app. CoolaData monitors the effects of every change you make in your app, such as a feature update, marketing campaign message or channel, by segmenting users based on their behavior. For instance, you may add a push notification (“Why haven’t you gone out on your first run yet…?”) that pops-up on the third day after install to retain and increase loyalty. The more users you have and the longer your app has been out, the more detail and variety of cohorts you’ll be able to analyze. This comparison (apples to apples) provides you with valuable information as to the effectiveness of your push notifications.
One of the most important metrics app developer track is user Lifetime Value (LTV), which is a dollar figure representing how much money the user spent or generated for you via ads. To calculate LTV when monetization is user-driven, we add up relevant actions, like app purchases, in-app purchases and exposure to page-views. Cohort analysis makes it possible to evaluate LTV by comparing different cohorts over a set period of time. A well-defined and tracked LTV is key to asses an app’s accurate state of affairs.
Revenue, retention, churn, ad ROI (the Return on Investment in advertising) – cohort analysis helps to monitor all these significant metrics. Analyzing different cohorts enables us to isolate and identify user reaction to a specific event – be it external, like a weather catastrophe, or something generated internally, like a marketing campaign or an app update. Actionable cohort analysis makes it easy to target relevant audiences and attend to their needs.
Now let’s get back to Ana and Mark. They are both avid runners – they run five (Ana) and four (Mark) times a week. And they are loyal RunApp users. You’d think that Ana and Mark would have a high LTV, spending on premium features, be they voice coaching, professional plans, or personal diets to match training routines. They are both in a cohort of users who run more than three times a week, but comparing Ana and Mark’s cohort to a cohort of those who run less than three times a week reveals that less frequent runners are actually the ones paying for premium features. In this case, cohort analysis disclosed the fact that the less professional runners are the ones generating more revenue. This granular inspection provides us with the relevant insights, funnels and paths we need to further delve into each segment, act on our insights and predict future trends.