I have wrote and spoken several times recently about how increasing user acquisition costs are impacting the social game industry, particularly social casino, and some strategies to offset and even capitalize on this trend. One of the best ways to combat this problem is to focus your resources on reducing churn rather than acquiring new users. As the cost of acquiring a user increases, the cost of replacing that user also increases. While companies devote ever larger budgets and build bigger teams for user acquisition, they fail to devote comparable resources to retaining the players they already have. Moreover, in the social casino space, players often play 3-5 apps, so the less they play your title, the more they are playing a competitor‘s slot.
I recently came across a blog post, “How to reduce Churn by building a bulletproof retention process” by Andrew Tate, that lays out five ways that you can reduce churn. Tate starts with the example of Burbn, a Foursquare clone that had a huge churn issue. Using analytics, the team saw that users found it too complicated and were using the app to post pictures rather than check in. So they modified the app based on this data to reduce churn and renamed it Instagram. Let’s just say it worked out okay for the Burbn team. This example shows how a focus on reducing churn can be the difference between failure and billions.
Focus on the small gains
The first step to combating churn is to take a lot of small steps rather than looking for the magic formula. One change or feature is not going to improve retention 50 percent. If you can reduce churn a few percent every build, you eventually will get negative churn rather than seeing your game lose all its users and revenue. Thus, it is more critical to build a process that always is working to reduce churn rather than fire off a churn reducing feature and then go on to your next KPI (key performance indicator).
To create the process discussed above, you first need to set goals. As with all goals, they need to be measurable and clearly defined as well as time based. Reducing churn is not a goal, but reducing churn by 50 percent in three months is a goal. Then you should break this goal into sub-goals, with Tate suggesting
- Reduce short-term churn by 20 percent
- Reduce medium-term churn by 20 percent
- Reduce long-term churn by 10 percent
With goals in place, you can track your progress as you create individual AB tests and set an end-point for the process. This technique aligns the entire team on a visible and understandable goal.
Ask the right questions
The next step is to ask the right questions of the right people. To get actionable insights, you need to use questions to understand specifics of why users are churning. You want to ask questions that address all of your churn goals, if they are the ones above, then you need to understand why people are churning both short term, medium term and long-term. Short-term churn is likely generated from different factors than long-term, for example. Also, different segments of customers (based on cohorts, usage patterns, etc) may be churning for different reasons.
Once you have the questions, there are many ways to answer them. First, observe your users. Look at the data they are generating. Talk to them. Also observe your competitors, what are they doing to reduce churn. Finally, ask others what they are doing to reduce churn.
Build a theory
Once you have requested, and collected, responses, questions you need to use the responses to create a workable hypothesis. As Tate writes, “The point of developing a hypothesis is that it forces you to think about the questions you’re asking in more detail. In particular, you have to think about the impact any experiment is going to have on churn. Hypothesizing the impact any experiment is going to have on your churn is going to help you with your decision making, turning the multitude of questions you have assembled into specific ideas you can action and test.”
Once you have created a hypothesis, you then prioritize ways to combat churn. These are prioritized the same as other features and experiments, the potential impact versus the cost (usually in development resources). Those with the greatest return on investment are put at the top of the queue.
Test and analyze
Once you have prioritized ideas for reducing churn, you need an implementation process. This should always done via experiments, either AB Testing or a multi-armed bandit approach. Release the new feature or make the change while keeping a control group. If the change has a positive impact (and not all will), then make it available to all users. Go through your list of prioritized features, testing each one and implementing those that reduce churn.
Refine, reject, repeat
Once you get the initial results from each experiment, you can implement it but you should also look to refine it. Maybe there is a further change that will magnify the impact. Even if an idea did not lead to a positive result, rather than discard it look to see if it might be more effective if implemented differently and then test again.
Reducing churn is not a one-time process. Do not run the experiments and move on to the next thing. Instead, keep refining the improvements you have made and keep trying new experiments to reduce churn further. It is like a guerrilla war; you are not going to win with a few major battles but continuously improving your position.
- With user acquisition costs increasing, reducing churn becomes increasingly important. The cost of replacing churned players is higher by definition when it costs more to get a new player.
- To reduce churn, you need to set specific goals, question why users are churning, build a hypothesis on how to combat the churn, then test, implement and refine.
- Reducing churn is not a one time event but a continuous activity. You always need to be looking at ways to reduce further churn by optimizing existing practices and coming up with new ones.