We’ve been testing different approaches to predict which users are about to churn for the past 6 months.
Finally built something that catches 70% of potential churners with minimal false positives. The prevention flows we trigger are converting at 40%.
Happy to share the signals we track and how we set up the automated responses.
Which prevention flows are you using? That 40% conversion rate is what counts. People get obsessed with prediction accuracy but ignore weak retention tactics. Catching 70% of churners? That’s good enough. The real value’s in automated responses that actually work. Are you hitting them with targeted offers based on usage patterns, or just running generic win-back campaigns?
What’s your churn window? We spent months spinning our wheels because we had the wrong timeframe.
We started at 30 days but switched to 14 - way better signal. If someone’s gone for 2 weeks, they’re probably not coming back on their own.
What prevention stuff are you running? Discounts or feature pushes? Usage reminders worked way better for us than cutting prices.
Nice work. What exactly triggers the prevention flows?
Those conversion rates look solid. Which signals worked best for predicting outcomes?
I’m tackling something similar but can’t nail the balance between accuracy and timing. How do you automate responses without coming off pushy?
What’s your false positive rate looking like? That’s always been our biggest headache.