The structural difference
| Health score | NPS | |
|---|---|---|
| Coverage | Every account, always | Accounts that responded |
| Source | Observed behavior in product, support, CRM | Self-reported answer to one question |
| Frequency | Continuous | Quarterly or annual campaign |
| Respondent | Not applicable | Whoever opened the email |
| Best use | Triage and renewal risk | Diagnosing why sentiment moved |
Why the respondent matters more than the score
The person who answers an NPS survey is usually an end user with a strong opinion about the interface. The person who cancels the contract is usually a finance or operations leader who never opened the survey and cares about a business result. A book full of promoters can still lose accounts when the buyer sees no case for the line item.
This is why sentiment data has to be weighted by role before it means anything. A detractor score from an economic buyer is a renewal risk. A detractor score from one power user who wants a keyboard shortcut is a product ticket.
How to combine them
Use the health score as the system of record for account risk and make NPS one low-weight component inside it, treated as missing data when the account did not respond. Scoring non-responders as neutral inflates the average and hides the accounts that went quiet.
Then use the free-text comments the way they are actually useful. They explain movement rather than predict it. When an account's health score falls two months running and the last survey comment named a missing capability, you have a specific reason to bring to the renewal conversation instead of a general concern.
Both numbers ultimately serve the same target. Retention shows up in net revenue retention, and the renewal forecast is only as good as the account-level risk feeding it, which means forecast accuracy depends on the score that covers every account rather than the one that covers the accounts willing to answer email.
Frequently Asked Questions
Is NPS a good predictor of churn?
On its own, no. NPS captures stated sentiment from a self-selected sample at one moment, and the person who answers is rarely the person who signs the renewal. Detractor scores are worth investigating, but a book of promoters can still churn when the economic buyer never responded to the survey and never saw value.
Can NPS be an input to a health score?
Yes, and that is its best use. Feed the response as one weighted component alongside usage and support signals, weighted low because coverage is partial. Treat a missing response as missing data rather than a neutral score, since scoring silence as average quietly inflates every account that ignored the survey.
What does a health score miss that NPS catches?
Intent that has not shown up in behavior yet. A customer whose team uses the product daily but whose leadership has already picked a replacement looks healthy on usage data. Open-ended survey comments and the sentiment behind them surface that decision earlier than the login curve does.
Which one should a customer success team run on?
Run operations on the health score and use survey responses as a diagnostic overlay. The health score covers every account continuously and drives the work queue. NPS covers a fraction of accounts periodically and tells you why the score moved.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like health score vs nps into prescriptive action for your team.
Schedule a Demo