What the survey doesn't capture
A customer can rate you nine out of ten and cancel six weeks later. Surveys measure attitude at a calm moment; cancellation happens at a moment of friction: a renewal with an unexpected increase, a poorly communicated claim, a charge that doesn't match what was promised. The useful signal is in behavior, not in opinion.
Behavioral signals that do anticipate churn
The models that work combine observable behavior variables with contact events.
- A drop in usage or interaction frequency relative to the customer's own historical pattern
- Contact with service without resolution on the first attempt
- Rejection or delay of the first charge after a change in terms
- Inquiries about cancellation or portability terms
- A switch in contact channel with no apparent explanation
Prioritize by value at risk, not by probability
A churn model ranks customers by probability of leaving. The retention decision must rank them by value at risk: probability multiplied by expected remaining value, minus the cost of the retention action. Retaining with a discount a customer who was leaving anyway destroys value twice.
Intervene where the decision happens
Effective retention doesn't happen in the month of cancellation but at the friction moment that triggers it. Redesigning the communication of a rate increase, moving up proactive contact after an incident, or reframing a renewal costs little and prevents most avoidable departures.
Key takeaways
- Observed behavior predicts churn far better than stated satisfaction.
- Prioritize by value at risk, not by probability of cancellation.
- The right intervention happens at the moment of friction, not at cancellation.
NS frameworks and reference sources
- Proprietary NS framework · DD·NS — methodological notes from the Behavioral Economics practice
- NS project base: anonymized cases by sector and geography
- Open market evidence and academic literature, recalibrated with client data
- DellaVigna & Linos (2022), 'RCTs to Scale: Evidence from Nudge Units', Econometrica
- Thaler & Sunstein — Nudge
- Kahneman & Tversky — Prospect Theory
This article develops proprietary NS Business Strategy frameworks, drawing on our project base and on public industry literature and studies cited above. Figures are reference ranges; each project is measured against the client's actual data.