Skip to content
Novandri.

NPS and Its Discontents: What the Score Does Not Say

A single number is easy to track and easy to misread. Read what NPS measures, a worked interpretation, and what to use alongside it.

Rachmadwipa Novandri5 min read

The Net Promoter Score reduces customer sentiment to one number between -100 and 100. That simplicity is why it appears on board decks everywhere, and why it is frequently over-interpreted. The score records an answer to one question; it does not explain why, and it is not a forecast of growth.

Direct answer: NPS measures the balance of respondents who would recommend a company to a friend or colleague, on an 0–10 scale, grouped into promoters, passives, and detractors. AAPOR's standards govern the survey behind it — population, wording, and sampling decide whether the number means anything. The boundary: the score does not measure actual referral behaviour, is sensitive to sample composition, and cannot tell you what to fix, which the voice-of-customer article argues for turning feedback into decisions.

A four-step flow: define the population, fix the wording, keep the sample steady, then pair with behaviour.
Conceptual: the survey design decides whether it means anything. The score does not measure actual referral behaviour, and it is sensitive to sampling.

What the score captures

A stated intention. Asking "how likely are you to recommend?" records a reported attitude, not an observed action, which the AAPOR standards require you to state as such.

A distribution summarised. Promoters minus detractors collapses a richer set of answers into one figure, which the email-metrics article warns about for aggregate proxies.

A snapshot of respondents. Who was asked and when determines the result, which the survey standards make central.

A worked interpretation

The following is hypothetical and illustrative.

A company reports an NPS of 40, up from 30 a year ago. Three checks decide what it means:

  1. Sample composition. Was the survey sent to all customers or to a segment? If only engaged customers responded, the score is biased, which AAPOR's standards require you to disclose. Suppose the response rate was 12% — the result describes respondents, not the base.
  2. What changed. Did the score rise because detractors decreased or promoters increased? The same net figure can hide opposite shifts, which the email-metrics article treats as a reason to report distributions.
  3. Behavioural check. Did referrals actually increase, as the referral article measures with rewards and holdouts? If stated intention rose but behaviour did not, the score is a weak proxy, as the attribution article warns about credit rules.

Three conclusions follow:

  • The number alone is not a result, which the evaluation-rubric article argues for concrete evidence.
  • The distribution and the sample matter more than the net, as the survey standards require.
  • **The follow-up question — why? — is where value lies, which the voice-of-customer article treats as turning feedback into decisions.

What to use alongside it

The open-ended reason. Ask why a respondent gave that score and group the answers by theme, as Voice of Customer: Turning Feedback into Decisions recommends.

Behavioural metrics. Retention, repeat purchase, and referral conversion are observed, which the retention and referral articles cover as outcomes rather than intentions.

Segment cuts. A single average hides a problem in one group, which the customer-segmentation article argues for real differences.

The counterargument — that NPS is useless — is too strong.

It is a simple, consistent signal for tracking direction over time within one organisation, provided its limits are stated, as the brand-awareness article argues for signals with a defined scope.

Keeping the survey honest

The value of a trend depends on the survey staying comparable. Three practices protect it:

Fix the wording. Changing the question, even slightly, breaks the series, which AAPOR's standards require for valid comparisons over time.

Fix the population.

Report the sample with the score. A number without a response rate and a population is hard to interpret, which the email-metrics article argues for denominators.

These practices are unglamorous and determine whether the score is a signal or noise, which the evaluation-rubric article treats as the difference between a metric and a measurement.

What NPS cannot do

It cannot forecast growth. Stated intention and actual behaviour differ, which the referral article treats as a central measurement problem.

It cannot tell you what to fix. That requires the reason and the behavioural data, as Voice of Customer: Turning Feedback into Decisions explains.

It cannot compare across cultures without care. Response styles differ by region, which the survey standards treat as a wording and context issue.

A decision test

  1. State the population and response rate with every score, as AAPOR's standards require.
  2. Report the distribution, not just the net, as the email-metrics article advises.
  3. Ask why and group answers by theme, as Voice of Customer: Turning Feedback into Decisions recommends.
  4. Compare with behaviour — retention and referrals, as the retention and referral articles cover.
  5. Cut by segment to find hidden problems, as the customer-segmentation article argues.
  6. Keep wording stable over time, which the survey standards require for trend validity.
  7. Treat the score as a signal, not a verdict, as the evaluation-rubric article advises.

For the feedback loop, see Voice of Customer; for the survey standard, see AAPOR.

References & Further Reading

  • American Association for Public Opinion Research, 2026. AAPOR — The professional association for public opinion and survey research; its standards and resources underpin the survey-design discussion.
  • Wikipedia contributors, 2026. Demand generation - Wikipedia — An overview of demand generation as a marketing discipline distinct from lead generation; secondary reference used with the lead-generation entry.