Net Promoter Score (NPS) Calculator

Net Promoter Score is reported as a single integer between -100 and +100. Customers answer one question, How likely is it that you would recommend us to a friend or colleague, on an 11-point 0 to 10 scale. The formula is straightforward: NPS = (percent promoters) minus (percent detractors). Passives are counted in the total but not in the numerator. If 60 percent of respondents are promoters, 25 percent are passives, and 15 percent are detractors, the score is 60 minus 15 equals 45. The bounds come from the math: a population of all promoters scores +100, all detractors scores -100. The passive percentage never appears directly in the formula but still pulls the score toward zero by inflating the denominator. This calculator lets you enter either the three bucket counts or a list of raw 0 to 10 scores, and it does the bucketing for you.

NPS was published by Fred Reichheld of Bain and Company, working with consultancy Satmetrix and Bain partner Rob Markey, in the December 2003 Harvard Business Review article titled The One Number You Need to Grow. The pitch was deliberately blunt: replace long, low-response customer-satisfaction surveys with one question that correlates with growth, and stop arguing about which 5-question scale or 25-question battery to use. Reichheld followed the article with the 2006 book The Ultimate Question, then a 2011 revision (The Ultimate Question 2.0) and 2021's Winning on Purpose with Darci Darnell and Maureen Burns. NPS spread quickly through SaaS, consumer banking, hospitality, and B2B, and is now used in some form by roughly two-thirds of the Fortune 1000. Academic critics (notably Timothy Keiningham and colleagues in 2007 and 2008 papers) have pushed back on the growth-prediction claim, but the metric stuck because it is simple to run, easy to benchmark, and tracks well enough at portfolio scale.

The bands are fixed by Reichheld's original definition and are not configurable: promoters score 9 or 10, passives score 7 or 8, and detractors score 0 through 6. The cutoff at 6 surprises people; intuitively a 6 or 7 sounds neutral, but Reichheld's empirical work showed that customers who answered 6 or lower were the ones most likely to churn, complain, or warn friends off the brand, while only 9s and 10s reliably drove referrals and repeat purchase. Promoters typically have a customer lifetime value 3 to 5 times that of detractors in B2B SaaS and consumer retail studies. The trade-off is that the bands are coarse: a customer who improves from a 6 to a 7 jumps from detractor to passive and moves the score, but a customer who improves from a 0 to a 6 changes nothing. That makes NPS noisy at the individual response level and only useful in aggregate.

Benchmarks vary widely by sector, but rough industry averages from Bain's Net Promoter System data and the Qualtrics XM Institute 2024 industry report give a useful reference. Software and tech sit around +40 (Apple consumer reportedly around +60, leading B2B SaaS firms above +50). Retail averages around +50, with category leaders like Costco and Trader Joe's above +75. Financial services sits around +30, with credit unions usually above the major banks. Telecom is the most punished category, often in the 0 to +20 range, with US wireless carriers historically near zero or negative. Airlines, cable, and health insurance also cluster low. Bain's broad bands classify any score below 0 as needing improvement, 0 to 29 as good, 30 to 69 as great, and 70 to 100 as world-class. Always compare yourself against your industry, not the absolute number; a +35 in telecom is exceptional, while a +35 in tech is mediocre.

NPS percentages are very sensitive to small samples because each response moves the score by a fixed amount: with 10 responses, one extra detractor changes NPS by 10 points; with 1,000 responses, by 0.1 points. Standard statistical guidance (Bain, Qualtrics, Medallia, SurveyMonkey) is to treat sample sizes below 100 as directional only because the 95 percent confidence interval can easily span 20 to 30 NPS points. At N = 100 the typical margin of error sits around plus or minus 10 points, at N = 400 around plus or minus 5, and at N = 1,000 around plus or minus 3. For benchmarking against competitors, comparing time periods, or steering executive decisions, target at least 400 responses per cohort. The 100-response cutoff in this calculator is a soft floor: above it the score is meaningful enough to act on, below it treat the result as a hint and gather more data before drawing conclusions.