A previous post focused on the how lifecycle marketing differs from live ops. Here we will dive deeper into lifecycle marketing. Specifically we will discuss how data analysis is used post launch to improve the native built-in onboarding flow, with the goal of improving retention.
Most games and apps include an onboarding tutorial that guides newbies around. But that is not all that is needed to retain users. To increase retention further, two other techniques can be attempted.Both rely on studying the differences between retained and churned users. Then we identify each user's lifecycle stage, and market to them appropriately.
We start by comparing the behavior of churned and retained users in the first two days. The technical steps are straightforward:
You should end up with a table that looks like figure 1 below. The first column shows whether the user churned or retained; the second column shows the event; and the last columns are the key KPIs.
Figure 1 - descriptive analysis of churned versus retained users, based on their early behavior.
Looking at the green highlighted events, we can see that more of the retained users had a "select outfit" event than the churned users. And more importantly, the frequency of the "select outfit" event per user for retained users is more than twice that of the churned users. The red highlighted event is an example where the frequency of the event occurrence is not different between the retained and churned segments.
Lifecycle marketing implies you should market to users who will churn; and not confuse the users who will be retained. So the first step is to figure out who will be retained and who will churn. Based on the simple descriptive analysis, you can now create heuristic segments in your CRM tool such as Leanplum. In the above example, a churn segment could be defined as
The marketing action to retain this user segment is dependent on the game capabilities. However we have seen simple nudges that educate users on the specific event work.
Another way to figure out which users will churn or retain, is to use predictive analytics. The general idea is the same, comparing churned to retained users. But rather than using simple event counts, predictive analytics uses all available event counts, player and device characteristics to build a predictive model. Then this predictive model is used to group active users into "likely to churn" or "likely to retain".
This approach is more comprehensive and responsive to changes to user behavior. The downside is the higher complexity and more moving parts. Check out Nick's video presentation at Casual Connect on our home page to learn more about predictive analytics.
Read more about how IMVU uses predictive analytics to identify the likely churn segment, and also to provide targeted rewards to these users. This case study describes not just the technical part, but rather how the churn predictions were used in an ongoing automated retention campaign.
Instead of analyzing the entire player base, start with looking at a recent week's worth of cohorts. In addition to reducing the amount of data processing required, the analysis will focus on the current app version and economic climate. What about possible seasonality? Once you have a usable framework with a week cohort, you can easily extend it to longer time frames.
To leverage any lifecycle analysis in a scalable fashion, games and apps should build an internal REST API to change user-level game configs, message users or provide rewards. May game-backend platforms provide such APIs, but it is cleaner to abstract out the tool specific API so that games have the flexibility to switch game-backends without affecting downstream dependencies.
Sonamine uses machine learning scores to execute live ops and lifecycle campaigns, contact us to learn more.
For a limited time until Mar 2024, Sonamine is offering a 60 day money back guarantee to Braze customers. Come experience the ease and simplicity of the First Time Spender Nudge package and watch your conversions soar.