Why use machine learning to compute health scores?

How are health scores calculated?

To calculate the health score of a customer, each event(which we collect using Intercom) is assigned a weight that reflects the impact it has on the likelihood churn, then multiply the weight by the event count, and sum them up and normalize. In many existing tools, the weights are determined by the gut. 😮

Use data, not your gut!

We love data. We don't have confidence using the gut to determine which events contribute the most for a customer to churn. We find that using machine learning to analyze data gives us the most accurate result. 🤖