Net Promoter Score (NPS) Calculator

Respondents rate 0–10. Scores of 9–10 are promoters, 7–8 are passives, and 0–6 are detractors.

NPS = % promoters − % detractors, reported as a whole number from −100 to +100. It is not a percentage, so write it as 40, not 40%.

The mistake worth avoiding: passives are excluded from the numerator but stay in the denominator. Dropping them entirely inflates your score.

The Net Promoter Score is a customer loyalty metric built on one question: "How likely is it that you would recommend [company] to a friend or colleague?", answered on a 0–10 scale. Fred Reichheld introduced it in the Harvard Business Review in 2003, and it was developed with Bain & Company and Satmetrix. Net Promoter, NPS and Net Promoter Score are registered trademarks of Bain & Company, NICE Systems and Fred Reichheld.

If you searched for an NPS calculator and wanted pension or retirement returns, this is the wrong page: that NPS is India's National Pension System, an unrelated government scheme. Everything below is about customer (and, with the same math, employee) Net Promoter Score.

Sort every answer into one of three groups: promoters (9 or 10), passives (7 or 8) and detractors (0 to 6). Then: NPS = (promoters ÷ total responses × 100) − (detractors ÷ total responses × 100). The result is a whole number between −100 (everyone a detractor) and +100 (everyone a promoter).

Worked example (illustrative numbers): 200 customers answer. 110 give a 9 or 10, 50 give a 7 or 8, and 40 give 0–6. Promoters are 110 ÷ 200 = 55%. Detractors are 40 ÷ 200 = 20%. NPS = 55 − 20 = 35.

Notice the 50 passives. They are not in either percentage, but they are in the 200 you divide by. If you drop them and divide by 150 instead, you get 73.3% − 26.7% = 47, a 12-point inflation from a spreadsheet mistake. Enter the same three counts in the calculator above and it returns 35 with a 95% interval of 24.0 to 46.0.

Not in a way anyone else can compare with. The promoter, passive and detractor bands are defined on the eleven-point 0–10 scale. Squeezing them onto five points (say, 5 = promoter, 4 = passive, 1–3 = detractor) produces a number with the same name and different behavior, and a 1–10 scale quietly removes the zero. If you already have five-point data, report it as a top-two-box satisfaction percentage instead (our CSAT calculator does that) and switch to 0–10 for the next wave.

An NPS is the difference of two proportions taken from the same sample, so its variance is not the familiar p(1 − p). Written with p as the promoter share and d as the detractor share (both as decimals), the variance of one response is p + d − (p − d)². The standard error of the NPS is the square root of that divided by n, times 100 to put it on the NPS scale. A 95% margin of error is 1.96 times the standard error.

In the worked example: p = 0.55 and d = 0.20, so the variance is 0.75 − 0.35² = 0.6275. Divide by 200 and take the square root: 0.056, or 5.6 NPS points. Times 1.96 gives a margin of error of about ±11. The honest way to report that survey is "NPS 35 (95% CI 24 to 46, n = 200)".

This is the interval the calculator shows. Brendan Rocks' 2016 paper in The American Statistician, which tested interval methods against 1,098 real NPS distributions from 236 companies, found this plain (Wald) interval covers the true score less often than it claims, especially at small samples. His best simple fix: add 0.75 to the promoter and detractor counts and 1.5 to the passive count, then run the same formula. For 15 promoters, 9 passives and 6 detractors (NPS 30), the plain interval is 2.1 to 57.9 and the adjusted one is 0.7 to 53.9. Above a few hundred responses the two barely differ.

To compare two independent surveys, combine their standard errors: the margin of error on the change is 1.96 × √(SE₁² + SE₂²). Illustration: last quarter 100 promoters, 60 passives and 40 detractors (NPS 30, SE 5.5); this quarter 108, 60 and 32 (NPS 38, SE 5.3). The margin on the 8-point rise is about ±15, so the data are consistent with no change at all. Run each quarter through the calculator and look at whether the intervals overlap heavily; if they do, hold the celebration.

Most "average NPS by industry" tables online do not say who was surveyed, when, or how the question was asked, and NPS moves a lot with method: a survey sent right after a support ticket, a relationship survey to all customers, and a panel of consumers rating brands they have used will give different numbers for the same company. Treat any benchmark without a methodology note as decoration.

One that does publish its method is Qualtrics XM Institute's 2024 US consumer study: 10,000 consumers rating 354 companies across 22 industries, with each industry's score the average of its companies' scores. Industry averages ran from 15.8 (car rental) to 34.3 (grocery). Those are consumer-panel scores for B2C brands; your own survey of your own customers is a different measurement, and B2B relationships are not covered at all.

The labels the calculator prints (below zero, needs work, good, great, world class at 0, 10, 30 and 50) are rules of thumb, not a benchmark. The comparison that reliably means something is your score against your own previous score, collected the same way, with the margin of error beside both.

NPS summarizes intent to recommend. It is a poor fit when recommending makes no sense (a tax authority, a monopoly utility, an internal IT tool), when you need to judge a single interaction (use CSAT or CES right after it), or when the sample is too small for the margin of error to be useful. Keiningham and colleagues' 2007 study in the Journal of Marketing found NPS was no better than other satisfaction and loyalty measures at predicting company growth, which is a reason to treat it as one signal, not the scorecard.

Our comparison of NPS vs CSAT vs CES covers which metric to use at which moment.

The score tells you how many; the follow-up tells you why. The standard follow-up is open text: "What is the main reason for your score?" Asking it of everyone gives you comparable reasons across the three groups. If you branch, keep it short: detractors get "What would it take to raise your score by one point?", promoters get "What do you value most?", and passives get "What is missing?".

Two practical rules. Keep the survey to the NPS question plus one or two follow-ups; every extra question cuts responses and widens the margin of error you just read about. And tag the open-text answers into a handful of themes before the next wave, so you can see whether the reasons behind the score change, not only the score.

In Zunoform, the NPS question type renders the 0–10 scale, can ask a built-in follow-up ("What is the main reason for your score?" by default), and can drive logic on the score, so detractors can be sent to their own follow-up. The form's summary tab shows the promoter, passive and detractor split and the NPS for you; paste those counts into this calculator to get the interval. Publishing a form with an NPS question needs the Pro plan; you can build and preview it on Free.

Several templates already include an NPS question: the customer exit survey, patient satisfaction survey, post-event feedback, onboarding feedback and churn survey. If you run NPS in Google Forms instead, our guide to NPS surveys in Google Forms shows the linear-scale setup and the Sheets formulas.