Why Your NPS Score Is Misleading You (And What to Measure Instead)

Every quarter, thousands of SaaS companies run the same ritual: send the "how likely are you to recommend us" survey, wait for scores to trickle in, then build a slide around whatever number comes back. It feels like measurement. Often, it's theater.

The number everyone quotes, and the number nobody checks

Net Promoter Score has been the default health metric for customer relationships since Fred Reichheld introduced it in 2003. It's simple to collect, easy to benchmark, and effortless to put on a dashboard. That simplicity is exactly the problem.

Research from the Marketing Science Institute, analyzing more than 250,000 customer ratings, found that NPS explains roughly 1% of the variance in actual customer spending. A company's NPS tells you almost nothing about whether a customer will buy more, renew, or walk. A single number, collected once or twice a year, is being asked to predict behavior it was never built to predict.

Why the score moves for reasons that have nothing to do with your product

NPS isn't just statistically weak, it's contextually unstable. Global benchmark data from Survicate's 2025 report shows median NPS varying from -1.21 in Croatia to over 9 in the Netherlands, a gap driven by cultural response styles (skepticism, precision, understatement), not service quality. A German customer and a Brazilian customer can have an identical experience and give wildly different scores.

Industry context distorts it further. A SaaS company scoring 40 might be beating its category by 20 points, or trailing best-in-class peers by 30, depending entirely on the benchmark you compare it to. Treating a raw NPS number as a universal signal, without adjusting for geography, industry, or survey timing, produces conclusions that don't hold up.

There's also a maturity paradox worth naming: teams with less developed customer intelligence stacks report leaning on NPS more heavily (around 43%) than teams with stronger, more integrated systems (around 30%). NPS tends to persist not because it's the best option, but because it's the only one many teams have.

NPS isn't useless, it's just late

To be fair to Reichheld's original framework, NPS does carry some signal. A declining score can precede rising churn by a quarter or two, functioning as a lagging-but-directionally-useful indicator. The mistake is treating it as sufficient on its own. Research on customer effort (the CEB study behind "Stop Trying to Delight Your Customers") found that reducing friction in a customer's experience predicts loyalty better than any single delight-driven score. Effort, usage, and revenue behavior tell you what's happening now. NPS tells you what someone felt when they filled out a form, maybe months ago.

What to measure instead: a composite customer health signal

Rather than replacing NPS with a single other metric, the more defensible approach for B2B SaaS teams is to stop relying on any single number and build a composite view instead. A practical framework:

1. Product usage and engagement — feature adoption, login frequency, depth of use by role. This is the closest thing to ground truth for whether a customer is getting value.
2. Customer Effort Score (CES) at key moments — onboarding, support resolution, renewal. Effort spikes are one of the earliest predictors of churn risk.
3. Net Revenue Retention (NRR) and expansion signals — are accounts growing, flat, or shrinking in spend? This is lagging but it's the metric that actually reflects business impact.
4. Support and success team qualitative signal — tickets, call notes, and CSM observations often surface risk weeks before it shows up in any survey.
5. NPS as one input, not the headline — keep it for longitudinal trend and benchmarking, but stop presenting it as a standalone verdict on customer health.

The teams getting this right aren't abandoning surveys. They're pulling usage data, support signal, and qualitative feedback into one place they can query in real time, instead of waiting for the next quarterly survey cycle to tell them what already happened. That shift, from static scorecards to a live, queryable view of the customer, is what separates teams that catch churn risk early from teams that find out in the renewal call.

FAQ

Is NPS completely useless?

No. It's a reasonable longitudinal trend indicator and an easy way to benchmark against peers over time. The problem is using it as a standalone measure of customer health or a predictor of revenue behavior, which the underlying data doesn't support.

What should replace NPS for B2B SaaS retention forecasting?

No single metric replaces it cleanly. A composite of product usage, Customer Effort Score at key moments, and Net Revenue Retention gives a far more reliable read on churn risk than NPS alone.

How often should customer health be reviewed if not through quarterly NPS surveys?

Usage and support signals can be reviewed continuously or monthly, since they update automatically. Save structured relationship surveys (NPS, CSAT) for quarterly or biannual cadence, as a complement, not the core system.

Does industry benchmark data make NPS comparisons meaningless?

Not meaningless, but they have to be read in context. Comparing your NPS to a same-industry, same-region benchmark is far more useful than comparing it to a generic "good NPS is 50+" rule of thumb, which ignores cultural and category variance.

---

Hugo Cusanno, founder of Personæ

---

Sources: Marketing Science Institute research via Lorikeet CX · Survicate, Global NPS Benchmark Report 2025 · CMSWire, NPS adoption trends · Armatis, NPS 2026 benchmarks