Net Promoter Score is the percentage of promoters (9–10 on the 0–10 recommend question) minus the percentage of detractors (0–6), a score from −100 to +100. Enter your three counts below to calculate NPS — along with the number every other NPS calculator leaves out: the margin of error that says how much of your score is signal.
NPS is calculated by sorting every response to “How likely are you to recommend us?” into one of three buckets: promoters rate 9 or 10, passives rate 7 or 8, detractors rate 0 through 6. The score is simply the promoter percentage minus the detractor percentage. Worked example: 200 responses — 100 promoters, 60 passives, 40 detractors. Promoters are 50%, detractors are 20%, so NPS = 50 − 20 = +30.
Two mechanics people get wrong: passives count in the total (they dilute both percentages — that is by design), and the score is written in points, not percent. An NPS of 30 is not “30%” of anything; it is a 30-point gap between your enthusiasts and your critics.
NPS is a difference between two percentages, and differences are noisier than either percentage alone — the statistical spread of an NPS is roughly double that of a single share. From 150 responses, a score of 37 carries a margin of error around ±12 points at 95% confidence: the true value is plausibly anywhere from the mid-20s to about 50. A dashboard that prints “NPS: 37” without that range is claiming precision the data does not have — which is why this NPS calculator prints the margin next to the score.
The practical consequences: month-over-month wiggles smaller than the margin are noise, celebrating a 4-point rise on 200 responses is celebrating a coin flip, and comparing two segments' scores needs a significance test, not a glance. This is the same discipline as any survey statistic — margins on honest bases, differences tested before they become headlines — applied to the one metric executives quote most and qualify least. (Our reports treat NPS like every other survey estimate: tested, with the margin printed.)
One more honesty note: the margin assumes a random sample of your customers. Real NPS response bias runs in a known direction: the delighted and the furious answer, and the quiet middle does not. That bias widens true uncertainty beyond anything the printed margin captures. Who was invited and how many answered matter as much as the score; customer feedback surveys, done properly covers that discipline.
The instrument is standardized: “How likely are you to recommend [company] to a friend or colleague?” on a 0–10 scale. What varies is timing — transactional NPS right after an interaction, relationship NPS on a periodic cycle — and mixing the two in one trend line muddies both.
NPS measures relationship loyalty, CSAT measures satisfaction with a specific experience, CES measures how much effort a task took. They answer different questions: use CSAT for transactions, CES for support flows, NPS for trajectory — not one score for everything.
The score tells you where you stand; the open-ended follow-up — “What is the main reason for your score?” — tells you why, once the verbatims are coded into themes. A tracked score without coded reasons is a thermometer without a diagnosis.
Question wording, transactional vs relationship timing, and the response-bias discipline are covered in depth in customer feedback surveys, done properly.
Count each response to the 0–10 recommend question into three groups: promoters (9–10), passives (7–8), detractors (0–6). NPS is the percentage of promoters minus the percentage of detractors, giving a score from −100 to +100. Example: 200 responses with 100 promoters (50%) and 40 detractors (20%) gives an NPS of 30. Passives do not enter the subtraction, but they do count in the denominator — leaving them out inflates both percentages.
Above zero means promoters outnumber detractors; above 30 is commonly considered strong and above 50 excellent in most consumer categories. But honest context matters more than any threshold: scores vary widely by industry, survey mode, and timing, so comparing your NPS to a published benchmark collected differently is close to meaningless. Your own trend, measured the same way each wave with the margin of error in view, is the comparison that means something.
Generally yes — 37 means promoters outnumber detractors by a comfortable margin, and it clears the +30 bar most practitioners call strong. The more useful question is the precision behind it: from 150 responses, an NPS of 37 carries a margin of error around ±12 points, so the true score is plausibly anywhere from the mid-20s to about 50. Treat the level as good, and treat movements smaller than the margin as noise.
For a margin of error near ±5 points at 95% confidence you typically need 900–1,500 responses, depending on your promoter/detractor mix; ±10 points needs roughly 250–400. Most teams tracking NPS on a few dozen responses per month are reading noise. If your volume is small, report the score with its margin, aggregate across longer periods, and judge direction over several waves rather than reacting to any single reading.
Related: sample size calculator · customer satisfaction survey software · the Likert scale, in depth
Brief the agent and field a real customer study — NPS with drivers, weighted, significance-tested, exportable. From $99 per study, no subscription.
Start a study