Net Promoter Score became a SaaS default because it's easy to measure, easy to trend, and easy to present to a board. What it doesn't do well: predict whether a specific account will renew.
The research on NPS as a retention predictor is underwhelming. Detractors (0-6) do churn at higher rates than promoters (9-10) on average. But the distribution is wide. Plenty of promoters churn. Some detractors renew for years. The individual account predictive power is weak.
Why NPS fails as a churn predictor:
Survey response bias. Accounts that respond to NPS surveys are not representative of your full base. Non-respondents churn at dramatically higher rates than respondents, and they're systematically excluded from your NPS calculation.
Recency bias. NPS responses reflect the most recent experience, not the long-term relationship. A customer who just had a great support interaction will score high even if they're planning to leave.
Role misalignment. The person who fills out the NPS survey is often not the person who makes the renewal decision. An enthusiastic end-user 9 doesn't protect you from a budget-cutting CFO.
What actually predicts individual account churn better than NPS:
Product engagement health scores (built on behavioral data, not surveys). Product usage patterns at 30/60/90 days post-purchase. Champion stability (tenure and role changes). Ticket volume trend (increasing tickets = increasing friction). Integration depth (more integrations = higher switching cost). Multi-team adoption (one team = single point of failure).
Use NPS for brand and aggregate sentiment tracking. Use behavioral signals for individual account churn prediction. These are different jobs requiring different tools.