Evan S., if you have enough historical data, tools like Mixpanel or Amplitude can help you identify the behaviors that most correlate with churn. From there, you can build a health score and refine it over time as you test its predictability.
Once you have a baseline, you can layer in more sophisticated signals: support data, email disengagement, decision-maker departures, key role hires, G2 category searches.
Depending on your product and engagement cycle, the predictability can get surprisingly good. At LANDR, we could detect 60% just based on segments. I’ve been told PandaDoc can predict 80% of churn ahead of time.
