This is how we do it at InnerTrends.
First, we split users into 4 categories:
- High engagement users
- Medium engagement users
- Low engagement users
- No engagement users
These categories are based on the users’ engagement scoring, which we calculate as follows:
- We create a list of engagement actions. These actions answer two essential questions:
- What is the reason people come to our product?
- What actions do they perform to reach their goals?
- Each time a user does an engagement action, they receive engagement points.
- Their engagement score is the result of their engagement frequency and engagement points.
- A data science algorithm then splits the users into 4 categories.
Second, armed with these figures, we can analyse and compare both types of behavior - of the ones engaged with an app and the ones who are not:
This allows us to understand what brings users back to the app, and
- for the highly-engaged users: what keeps them into the app
- for the unengaged users: why do they keep coming back to the app but not perform any engagement actions
We use this data to target the users based on their engagement level and decrease as much as possible the number of users who are not engaged. It’s much easier to do it like this than target users who have already churned.