Why Your CS Health Score Is Wrong 40% of the Time (And How to Fix It)
Rules-based health scores misclassify 30-40% of at-risk accounts. Here is the math, the reason, and the operational fix.
Tactical predictive analytics content for Customer Success, RevOps, and Data teams. Written by Ronen Mesonzhnik.
Rules-based health scores misclassify 30-40% of at-risk accounts. Here is the math, the reason, and the operational fix.
Most teams assume building a churn prediction model takes months. Here is what the actual timeline looks like when you remove the SQL, the data science team, and the implementation project.
Most churn models end up as a dashboard that nobody checks. Here is how to wire churn probability scores directly into HubSpot so CSM tasks create themselves before the account goes dark.
Vendors claim their AI churn model was trained on millions of SaaS accounts. Here is why that pitch sounds better than it actually performs - and what your model needs to include instead.
In February 2026, nearly $1 trillion in software market value evaporated in a week. The headline blamed AI. The real story is simpler and worse: SaaS retention is breaking, acquisition-led growth is over, and most companies still cannot see churn coming. Here is what the data actually says, and the only operational answer that holds up.
Mid-market SaaS companies evaluating predictive analytics keep landing in the same three-way funnel: Pecan, DataRobot, or a no-code tool. Each is built for a different team and a different bar of technical skill. A sourced 2026 comparison.
If you run Customer Success or RevOps without a data scientist, here is the practical sequence: what to predict first, what data you actually need, and what to do with the predictions once you have them.
Most CS teams build health scores from one signal: usage. The signals that actually predict churn 60-90 days out are sitting in HubSpot, untouched. Here are five of them.