Mobile App Retention – What Reduces Your App Churn?

Written by: on November 9, 2016

How to efficiently optimize your app for customer retention.

According to leading app analytics firm Localytics, about 1 in 4 new app users will abandon an app after a single launch. In fact, 2 out of 3 users will have deleted an app before their 11th session. Clearly, retaining users is becoming increasingly difficult in an environment where tens of thousands of apps compete for time and space in every possible category. So how can your app maintain its prominent place on your customers’ devices?

You can find articles all over the internet promoting strategies for creating “sticky” apps. Gamification, push notifications and social features are just a few worth considering. While many of them are very likely to encourage retention, it is impossible to compare their value among your unique audience just by reading. At POSSIBLE, we lean heavily on behavior data to support feature development priorities. By examining user data in a thoughtful way, we can be more cost effective when allocating limited resources to optimize an app. But first things first: Thoughtful analysis is next to impossible without a carefully planned analytics implementation, so let us start there.

The most important analytics consideration when trying to plan for retention optimization is to take the time to build a robust and intuitive measurement strategy. As part of the process, you should identify and tag only the key events that make an impact to your business and user experience. Over-tagging an application can make it difficult to discover and prioritize opportunities given the amount of ‘noise’ in your data. Some examples of areas we typically focus on are video milestones, process completions, information capture, bookings and achievements. Each event should be tracked within an appropriate hierarchy and/or contextual data model. This enables the flexibility to create intuitive ad-hoc behavior segments that might correlate with higher retention. While some analytics vendors offer far more in the way of custom segmentation capability, all the major platforms should meet your needs. POSSIBLE Mobile is agnostic when it comes to analytics vendors, but the vast majority of our clients use either Google Analytics, Adobe, or Localytics.

Now that you have all your event and content tracking in place, let’s take a deeper look at how we might assess user behavior in an app to optimize retention: After monitoring traffic, content and feature usage patterns, you should have a good understanding of where people are going, and what they are doing most. The next step is to build behavior segments in your analytics platform that are based on highly used features relative to their exposure. For example, group A ‘responded to X alerts’ or group B ‘watched >3 videos’. The goal is to discover wide behavior variations within these groups, but more specifically, which ones correlate highly with increased retention rates. Do visitors who have a greater propensity to leverage certain app features end up staying longer? Do they do more of what we want them to do relative to others? Are they visiting more frequently over a longer time span?

As we begin to see high correlations between certain features and retention, we need to take our analysis a step further. Our goal is to prove causation within our correlations. In other words; do they return more often as a result of this feature or do those that stay just happen to use more of it. In an ideal scenario this involves a basic A/B testing tool such as Optimizely or Leanplum. If we wanted to prove that push notifications cause retention, for example, we would divide our audience into ‘test’ and ‘control’ groups. Our test group (let’s say 5% of all new users, to minimize impact) launches an app with notification settings completely hidden. Our control group maintains all of the available features. Now let’s compare. Did the test group have better, similar or worse retention over time? If the delta is large and statistically significant, it may be a signal to invest more in push. But we shouldn’t stop there. In order to invest wisely, we need to understand the relative impact of push to other features and strategies. What if another feature is far more useful to users in the long term? Perhaps it is significantly cheaper to test and develop. Without further testing, you risk leaving money on the table. We suggest testing 2-4 features that compete for resources. Once changes are made, you may want to get more granular with that particular feature, or test something new. Win or lose, every test should be considered a new opportunity to learn.

In many of our recent experiences, we’ve found that investments in mobile app analytics and optimization have not kept pace with the growth of the medium. Often times, feature development is based on subjective opinion or anecdotal evidence. As a result, there is ample opportunity to remain ‘sticky’ in today’s high-churn app marketplace by using behavior data to drive retention strategy.

If you would like to hear more about how our analytics team can increase return on your app investments, don’t hesitate to reach out.

Brad Gagne

Brad Gagne

Brad Gagne has been a leader in the POSSIBLE analytics group for over 10 years. Most recently, he was Group Director of Marketing Sciences for the Seattle office, and recently shifted roles to focus on growing analytics for mobile and connected devices at POSSIBLE Mobile. His current and past clients include Turner Broadcasting, Microsoft, TDAmeritrade, IHG, Hyundai and several others.  

Add your voice to the discussion: