Net Promoter Score calculation: formula, example and analysis worksheet

Net Promoter Score (NPS) is the percentage of respondents who give a 9 or 10 on the standard 0–10 recommendation item minus the percentage who give a 0 through 6. Scores of 7 or 8 are passives: they remain in the denominator but are not added to either side of the subtraction. Report the score with its respondent count, field period, population and uncertainty—not as a percentage or an automatic prediction of growth.

What is Net Promoter Score?

Net Promoter Score is a customer-loyalty metric based on a 0–10 question about likelihood to recommend an organization, product or service. Bain describes the broader Net Promoter System as a loyalty-management approach. The calculation itself is a compact summary of the response distribution, not a complete measurement system. See Bain's current NPS overview.

What is the NPS formula?

NPS = percentage of promoters − percentage of detractors. Use all valid 0–10 answers as the denominator.

The possible result runs from −100, when every valid respondent is a detractor, to +100, when every valid respondent is a promoter. Because two percentages are subtracted, write an NPS of 35 as 35, not 35%.

Worked example

A survey receives 200 valid 0–10 answers: 110 promoters, 50 passives and 40 detractors. Promoters are 110 ÷ 200 = 55%. Detractors are 40 ÷ 200 = 20%. The result is 55 − 20 = NPS 35. Passives do not appear in the final subtraction, but their 50 responses stay in the denominator.

Reusable calculation worksheet

How should uncertainty be reported?

NPS is a statistic estimated from a sample. Its sampling error depends on the shares of promoters and detractors, their covariance and the sampling design. For a simple unweighted random-sample approximation, let p_P and p_D be the promoter and detractor proportions. An approximate standard error is 100 × √[(p_P + p_D − (p_P − p_D)²) ÷ n].

In the worked example, p_P=0.55, p_D=0.20 and n=200. The approximate standard error is 5.60 NPS points; a normal 95% interval is roughly 35 ± 11.0, or 24.0 to 46.0. This approximation is not valid for every design. Use survey-analysis software for weights, clustering or stratification, and avoid treating overlapping point estimates as proof of a real change.

How should NPS changes be interpreted?

1. Confirm the population, invitation method, question wording, placement and field period are comparable.
2. Compare the full 0–10 distribution and category counts, not only the net score.
3. Calculate uncertainty using the actual sample design.
4. Review missingness, response rate and any changes in customer mix.
5. Analyze the open-ended reason with a documented coding method.
6. Pair the score with behavioral and operational measures relevant to the decision.

Worked sensitivity check

Suppose the next wave has 56% promoters and 19% detractors, giving NPS 37. The two-point increase is not automatically meaningful. If the samples are small or composition changed, the uncertainty interval may easily include no change. Publish the counts and method before celebrating the direction.

Transparent method

We verified the conventional classification and formula against current first-party survey and NPS sources, then added a reproducible denominator check, uncertainty approximation and change-audit worksheet. The article does not rank software, imply that one metric predicts every business outcome or treat vendor examples as independent validation.

Limitations

NPS compresses an 11-point response distribution into three groups, so different distributions can produce the same score. Cultural response styles, sampling, nonresponse, question context and relationship stage can affect the result. Nielsen Norman Group cautions that NPS does not capture usability on its own and should be paired with behavioral measures. See the methodological discussion. Do not infer causation, future revenue or individual behavior from a score without separate evidence.