ChurnRX vs NPS
NPS told you who’s happy. ChurnRX tells you who’ll stay.
Net Promoter Score measures sentiment — and sentiment correlates about 0.01 with whether a customer actually stays. ChurnRX measures behavior, and turns it into a number you can act on.
The core argument
Satisfaction isn't fit.
The most-tested finding in churn research is also the most ignored: there is essentially no correlation between satisfaction and retention. NPS correlates roughly 0.01 with customer lifespan. Happy customers churn. Customers who log support tickets often stay longer, not shorter. Satisfaction and survival are nearly independent.
The reason is structural. A survey captures how someone feels in the moment they answer — filtered through who bothered to respond, how the question was framed, and what mood they were in. What actually predicts retention is whether customers achieve results with your product. So ChurnRX measures the behavior, not the opinion.
ChurnRX fits a survival model to your subscription data and computes your PMF Score — the share of customers who have bonded so completely they effectively never churn. It is the floor your retention curve flattens to: a structural fact about your base, not a sentiment reading.
Correlation between NPS and customer lifespan. A survey can tell you how people feel. It can’t tell you who’ll stay.
Side by side
Sentiment vs. behavior.
| NPS | ChurnRX | |
|---|---|---|
| Signal | Self-reported sentiment — how a respondent says they feel today | Observed behavior — whether customers actually stay |
| Output | A −100 to +100 score from a single survey question | A measured PMF Score — the fraction permanently bonded to your product |
| Predicts retention? | Barely — sentiment correlates ~0.01 with who stays | Directly — it is derived from retention itself |
| Comparable across companies | No — response bias and scale norms vary by audience | Yes — the shape of the survival curve carries no sales-driven distortion |
| Tells you what to do | No — a number with no map to the moves that change it | Yes — ranked retention factors, your real ICP, and where customers fail |
A fair reading
What NPS gets right.
NPS is cheap, fast, and universally understood — a single question that anyone can run and any board already recognizes. As a pulse on perception, it’s genuinely useful: it surfaces a sudden drop in goodwill, opens a verbatim feedback channel, and gives support and product teams a recurring read on how a release landed.
Used for what it is — a sentiment thermometer — it does its job. The problem isn’t the survey. It’s asking the survey to answer a question it was never built to answer.
- It predicts retention barely better than chance — sentiment and survival are nearly independent.
- It's not comparable across companies: response bias and scale norms differ by audience, so the same number means different things.
- It tells you nothing about what to do — there's no map from the score to the moves that would change it.
- It's distorted by who responds, when, and how the question is framed — a sample of feelings, not a measure of behavior.
Keep NPS for what it’s good at: a read on how customers feel. Use ChurnRX for the question it can’t answer — whether your growth is durable, and who is actually going to stay.
Stop guessing from sentiment. Measure who'll stay.
Upload a CSV and get your PMF Score — the behavioral measure of your product-market fit, benchmarked against the field.
Run your free benchmark