Net Promoter Score Calculator
Promoters minus detractors — with the 95% confidence interval around the result, because NPS is measured on a sample and on a few hundred responses that interval is wide enough to swallow most of the movement people report as progress.
Free · No signup · Runs entirely in your browser
Promoters minus detractors, as every net promoter score calculator does it — and then the part they leave out. NPS is measured on a sample, so it carries a margin of error, and on a few hundred responses that margin is wide enough to swallow most of the quarterly movement people report as progress.
Rated 9 or 10.
Rated 7 or 8. Counted, never scored.
Rated 0 to 6.
| Group | Rating | Responses | Share |
|---|---|---|---|
| Promoters | 9–10 | 50 | 50.0% |
| Passives | 7–8 | 30 | 30.0% |
| Detractors | 0–6 | 20 | 20.0% |
| Total responses | 100 | ||
Your score is 30, and on 100 responses the true figure is somewhere between 15 and 45. That is a span of 31 points. Any quarterly change smaller than about 21 points is indistinguishable from having surveyed a different set of customers — which is worth knowing before it goes in a board deck as a trend.
Note what the passives column does to the score: nothing. A company with 30.0% passives and one with none can post identical scores while being entirely different businesses — 50 promoters against 50 detractors scores zero, and so does a room of a hundred people who feel nothing at all. The score cannot tell them apart, so read the three shares alongside it rather than instead of it.
Scores move when the product moves. If you are building largely on your own, Tekk turns what you want into specs your coding agent can actually execute.
The calculation
NPS = (promoters ÷ total × 100) − (detractors ÷ total × 100)
Promoters rate 9–10, passives 7–8, detractors 0–6. Passives appear in the total and nowhere else.
Worked example
| Group | Rating | Responses | Share |
|---|---|---|---|
| Promoters | 9–10 | 50 | 50.0% |
| Passives | 7–8 | 30 | 30.0% |
| Detractors | 0–6 | 20 | 20.0% |
| Total | 100 |
50% − 20% = NPS of 30. Every calculator agrees to here.
The interval nobody prints
NPS is estimated from a sample, so it has a sampling error like any other statistic. Because it is a difference between two proportions drawn from the same sample, the two shares covary and the variance is not simply the sum of two binomial variances:
Var(NPS) = [p_promoters + p_detractors − (p_promoters − p_detractors)²] ÷ n
On the numbers above that is (0.5 + 0.2 − 0.09) ÷ 100 = 0.0061, a standard error of 7.81 points, and a 95% interval of:
| Score | 30 |
| 95% interval | 15 to 45 |
| Margin | ±15.3 points |
The honest statement is not "our NPS is 30". It is "our NPS is somewhere between 15 and 45, and 30 is the middle of that".
Which makes most quarterly movement noise
This is the practical consequence, and it is worth being blunt about.
| Responses per quarter | Is a move from 30 to 35 real? |
|---|---|
| 100 | No — far inside the noise |
| 1,600 | Still no — about 1.84 standard errors, just under the bar |
| ~1,800 | Yes, marginally |
A five-point improvement on a hundred responses is not an improvement. It is what happens when you survey a different set of customers. It gets reported as a trend in board decks every quarter, in both directions, and the underlying score may not have moved at all.
Getting to a ±5 point interval takes around 940 responses. Getting to ±3 takes around 2,600, because halving the margin costs four times the sample. Precision here is expensive, and the reasonable response is usually to treat the score as a band rather than to chase the decimal.
What the score cannot see
Passives are invisible. These two companies post identical scores of zero:
| Promoters | Passives | Detractors | NPS | |
|---|---|---|---|---|
| Polarising | 50 | 0 | 50 | 0 |
| Indifferent | 0 | 100 | 0 | 0 |
One has fifty people who would recommend you and fifty actively warning others away. The other has a hundred people with no opinion. These are entirely different problems with entirely different fixes, and the score cannot distinguish them.
The bands are conventions. Above 0 good, above 30 great, above 70 excellent — widely quoted, and not derived from anything much. Industry benchmarks vary far more than the bands suggest, and your own score last quarter is a more useful comparison than any published average.
It is one question. That is its strength and its ceiling. It tells you the direction of sentiment and nothing whatsoever about why, which is what the free-text box after it is for.
How it works
- 1
Enter your three response counts
Promoters rated 9 or 10, passives 7 or 8, detractors anything from 0 to 6. Every net promoter score calculator needs the same three numbers; what differs is what it does with them afterwards.
- 2
Read the range, not just the score
A hundred responses split 50/30/20 gives a score of 30 — and a 95% interval running from 15 to 45. The score is a point estimate drawn from a sample, and quoting it alone implies a precision the survey never had.
- 3
Check how many responses you actually need
The tool works backwards from your current mix to the sample size that would put the interval at ±5 points. On these defaults that is 938 responses. Most teams are reporting from a fraction of it.
Frequently asked questions
- What is Net Promoter Score?
- A single-question survey measure: how likely are you to recommend us, on a scale of 0 to 10. Respondents scoring 9–10 are promoters, 7–8 passives, and 0–6 detractors. It was introduced by Fred Reichheld in 2003 and has become the default customer-satisfaction metric in most companies.
- How do you calculate NPS?
- Subtract the percentage of detractors from the percentage of promoters. With 50 promoters, 30 passives and 20 detractors out of 100 responses, that is 50% - 20% = a score of 30. A net promoter score calculator does exactly this and no more; the result runs from -100 to +100, and it is a number rather than a percentage despite being built from two of them.
- How are passives treated?
- They count in the denominator and nowhere else — they dilute both percentages without adding to or subtracting from the score. That is a real blind spot: 50 promoters against 50 detractors scores zero, and so does a room of a hundred people who feel nothing at all. Read the three shares alongside the score, never instead of it.
- What is a good NPS?
- The conventional bands put anything above 0 as good, above 30 as great, and above 70 as excellent. They are widely quoted and not especially load-bearing. A net promoter score calculator can tell you which band you land in, but benchmarks vary enormously by industry, and your own figure from last quarter is more informative than any published average.
- Why does this nps score calculator show a confidence interval?
- Because NPS is a sample statistic and has a sampling error like any other. An nps score calculator that prints 30 and stops implies you have measured the number exactly. You have not — you have estimated it from whoever happened to respond, and on 100 responses the true figure could reasonably be anywhere from 15 to 45.
- How is the margin of error calculated?
- NPS is a difference between two proportions drawn from the same sample, so its variance is [promoter share + detractor share − score²] ÷ n, not simply the sum of two binomial variances — the two shares covary. The 95% interval is 1.96 standard errors either side, which on the default numbers is ±15.3 points.
- Is a 5-point change in NPS meaningful?
- Almost never at typical sample sizes. Moving from 30 to 35 on 100 responses per quarter is well inside the noise. It is still inside the noise at 1,600 responses per quarter — that comparison lands at about 1.84 standard errors, just under the 1.96 bar. You need roughly 1,800 responses in each period before a 5-point move is distinguishable from chance.
- How many responses do I need?
- For a ±5 point interval at a typical mix, around 940. For ±3, around 2,600. The relationship is quadratic — halving the margin costs four times the responses — which is why chasing a precise NPS gets expensive quickly, and why most teams should treat their score as a broad band rather than a figure.
- What is the net promoter score formula?
- The net promoter score formula is % promoters − % detractors, where each percentage is over all respondents including passives. Written out: (promoters ÷ total × 100) − (detractors ÷ total × 100). Nothing more — the apparent sophistication of the metric is entirely in how it is interpreted, not how it is computed.
- Should I use the net promoter score formula on a small sample?
- You can, but report the range with it. The formula does not care how many responses you have, which is exactly the problem: it returns a confident-looking number from twenty responses and from twenty thousand. Below roughly 100 responses the interval is wide enough that a net promoter score calculator is telling you very little on its own.
- What are the criticisms of NPS?
- That the 0–10 scale is chopped into three groups arbitrarily, throwing away information; that passives are ignored; that it is a poor predictor of actual growth despite the name; and that it is reported without error bars almost universally. The last one is the most fixable, which is why this tool leads with it.
- Do you store the numbers I enter?
- No. The calculation runs entirely in your browser. Nothing is sent to a server, nothing is logged, and there is no account.
- Why is this free, and what is Tekk?
- Tekk is a spec-driven development platform for people building software with AI coding agents. This calculator costs us nothing to run, and satisfaction scores move when the product moves. No signup, no run limit, no upsell inside the tool.
Want a real spec for what you’re building?
Drop a sentence. Tekk grounds it in your actual code and turns it into an executable plan.
Free to try · Connect GitHub during signup