How to use it
- Enter the number of respondents who gave each score from 0 to 10, or the totals of promoters, passives and detractors.
- Choose the confidence level for the margin of error, usually 95%.
- Read the NPS, the shares of each group and the margin of error.
- Compare the score with an earlier wave only if the difference is larger than the uncertainty around both scores.
Formula and logic
The score itself is simple:
NPS = % promoters (9–10) − % detractors (0–6)
Passives, who answer 7 or 8, count towards the total number of respondents but do not enter the subtraction.
The margin of error follows from the fact that each answer contributes +1, 0 or −1 to the score:
e ≈ z · √((P + D − (P − D)²) / n)
Here P and D are the shares of promoters and detractors as decimals, n is the number of respondents and z is 1.96 at 95% confidence. The result is multiplied by 100 to express it in NPS points.
Example: 200 respondents, of whom 100 are promoters (P = 0.50) and 40 are detractors (D = 0.20). The NPS is 50 − 20 = +30, and the margin of error is about ±10.8 points, so the true score most likely lies between about +19 and +41.
How to read the result
That example shows why NPS needs larger samples than a simple percentage: because it is the difference between two shares, its margin of error is wider. A change of five points between two waves of 200 respondents is almost always noise.
There is no universal "good" score. Results differ widely by sector, country and survey method, and scores collected straight after a purchase can differ markedly from those in a relationship survey sent months later. The most useful benchmark is your own trend, measured with the same question, timing and sample source.
The number alone does not tell you what to fix. Always add an open follow-up question asking why the customer gave that score, and code the answers into themes. Many programmes pair NPS with CSAT for individual interactions and CES for service effort. For a full programme, see our customer satisfaction and NPS research.
Frequently asked questions
What is a good NPS?
Any positive score means you have more promoters than detractors, but whether a particular score is strong depends on your sector, your market and how the survey was run. Comparing your score with figures published by other companies is risky, because their sampling and timing are rarely known. Track your own score over time and compare segments, products or branches measured in the same way.
How many responses do I need for a reliable NPS?
More than for an ordinary percentage. With 200 responses, the margin of error is typically around ±10 points, and with 400 around ±8, depending on the mix of promoters and detractors. If you want to compare branches or segments, each one needs its own adequate base. Enter your expected shares in the calculator to see the margin for your sample.
Why are passives left out of the formula?
They are not left out of the base. Passives count in the total number of respondents, so a larger share of passives lowers both percentages and pulls the score towards zero. They are excluded from the subtraction because the method treats scores of 7 and 8 as satisfied but not enthusiastic, unlikely either to recommend actively or to warn others off.
Is NPS better than customer satisfaction?
They measure different things. NPS captures willingness to recommend and works well as a single headline indicator for a relationship. Satisfaction scores are better at pinpointing how a specific delivery, call or visit went. Most programmes use both, with NPS at company level and satisfaction or effort scores at touchpoint level. If you want help designing such a programme, request a proposal.
Related reports
Customer satisfaction & NPS
Customer satisfaction research measures how well your products and service meet customers' expectations, usually with…
Read →Customer journey research
Customer journey research reconstructs the path people take from first noticing a need to buying, using and renewing,…
Read →Brand health tracking
Brand health tracking is a repeated survey that measures how many people know your brand, consider it, prefer it and…
Read →