To calculate the boundaries of these categories, the secret lies in starting from the engagement scores of customers who churned in a specific period of time. You have to determine the highest score of a person who churned.
- No engagement users are those who had no engagement actions during the last 30 days, though they used your app.
- To find out who your low engagement users are, eliminate outliers from the churned users (the atypical users). Basically, all the people who have a similar score to those who churned in the past will be part of your low engagement users.
- After eliminating your low and no engagement users you are left with your high and medium engagement users.
- Repeat the same process. Eliminate the atypical users, and you will have your high engagement ones.
- And then you are left with the medium engagement users.
3. Then, calculate the churn risk for each category.
To calculate the risk for each category, you need to:
- Select a period previous to 30 days ago
- Identify all the active customers from that period and the ones that churned.
- Identify the engagement category of each customer
- Calculate churn risk per category
For instance, to calculate the churn risk for high engagement customers use the following formula:
Churn Risk = Churned High Engagement Users / Active High Engagement Users
4. Decrease churn by targeting customers based on their engagement level changes.
The next step is to determine when users move from one category to another. And, believe me, they move a lot!
For instance for a company with 300 active users / month, about 10 of them will change category every day. For companies with thousands or tens of thousands of users, you can expect tens or even hundreds of users to change their engagement level every single day.
You definitely want to monitor the decreasing engagement scores, because someone who goes from high to medium and then to low engagement will have a very high churn risk. But, because they are still active, you can still interact with them, you don’t need to recover them after they left. The best thing to do is to find out the problem and fix it, and thus prevent them from leaving.
Preventing churn is an issue we are going to talk about in more detail in our next newsletter.
Till then, if you are interested in more details about Engagement Scoring & Churn Prediction
, don’t forget to check out the webinar
from last week.