We all understand the importance of retaining the users we acquired especially as our marketing and sales costs increase. Good mobile app retention rates can even be the crucial difference in startups that succeed and startups that fail. Bad retention can kill your startup so let’s figure out how to fix it.
Let’s start by defining what I mean by retention so we are all on the same page. To do that, we will use the metrics framework created by Dave Mclure which you can see in the image below:
Retention is directly affected by two other areas: Acquisition and Activation. We will focus on the activities that follow once you activate a user i.e. how to bring back users once they perform some crucial task within your app.
Let’s also separate retention from engagement, two areas that get easily confused. Andrew Chen, who works for Uber, defines retention and engagement in the following way:
“I generally define retention as simply the act of getting users BACK to revisit, regardless of their actual activity on the site. Contrast this with engagement, which measures how much time they spend with the product, how many features they interact with, etc.” – Andrew Chen
Pretty clear right? Before we get to the actual tools and techniques of improving retention, let’s answer the question almost every startup wants to know: what is a good retention benchmark (or am I doing better than my competitors)?
What Are Good Mobile App Retention Rates?
Benchmarks are useful because it can help us figure out when we are doing “good enough”. If the average retention rate in your industry is 5% and you are doing 10%, it makes sense to focus your efforts on other areas.
Mobile apps have pretty crappy retention rates and some reports have shown that losing up to 80% of your users is normal though the best apps tend to do better. One thing you will quickly realize is that everyone talks about retention in a different way.
Here’s a few example of how companies talk about retention:
- “Our D4 retention is 30%”. which means that 4 days after users install the app, 40% of those users are active on that specific day (day 4).
- “We retain 30% of users within 90 days” which means that 30% of users are still using the app 90 days after they signed up.
The action that qualifies you as “being retained” also differs for each app. Is opening the app enough to qualify as “retained”? Or do you have to perform some key actions within your app? Figure out what is important to your app and then focus on looking at the users who are completing those actions.
Another key factor in understanding retention is the type of app. Mobile games might see high retention rates in the first few days with a sharp drop off while food ordering apps might see a high mobile app retention rate across 90 days but very low daily usage (used only once a week).
If you aren’t tracking retention right now, I recommend looking at the overall retention over 90 days. Flurry (mobile analytics software acquired by Yahoo) created a chart showing some averages by different categories which you can see below:
Now that you know where you stand in comparison to other companies in your industry, we can move on to figuring out how to track and improve retention. While there are different techniques and reports for doing this, we will focus on the most versatile of all: cohort analysis.
Why You Should Be Using Cohort Analysis to Measure Your Retention
Cohort analysis is a way of grouping your users by a common characteristic such as sign up date. We can then see how this group of users interacts with our app over time.
Imagine that you’re trying to improve your conversion rate towards a paying subscription. It can be hard to control for the different factors that could affect this conversion rate, factors such as:
- Older users vs new users (older users may understand your app better)
- Traffic source
- Different pricing plans
You can solve this by grouping users into cohorts. You could see how the users who signed up for your product on June 2015 convert versus the users who signed up on June 2016. You could further segment down into each cohort for other factors like traffic source and demographics.
Ok Ruben, cohort analysis sounds pretty great but also complicated to set up. How do I get started? Lucky for us, most analytics tools make this really easy to do. For example, Mixpanel and Kissmetrics have a tool that lets you quickly create a cohort table based on different events.
The above image shows how a cohort table looks in Mixpanel. We are looking at users who completed Action A and then came back and completed Action B. Action A can be when the user signs up while Action B can be an important action in your product like uploading a photo.
We then divide our report by a time period (weekly in this example) and we can see that on the week of March 14, 11,722 users completed Action A (sign up) and then 65.45% of those users came back and did Action B within 1 week of Action A.
We can then see what percentage of users came back and did Action B in week 2, week 3, and so on. You can see that the second number gets smaller with each week as the app loses more users (churn).
You could then see how your changes to your app affect future cohorts (in this case weekly cohorts). In our example, the numbers get slightly worse but ideally you are able to increase the percentage of users that you retain.
Cohort analysis will be good enough for 90% of startup retention questions which makes it great technique to start using on regular basis. As your app grows, you can look into more advanced techniques for measuring retention such as cycle plots but that’s a lesson for another day.
I would love to hear your questions or comments. You can reach me at Twitter @ugarterd