eNPS Calculator: Your Employee Net Promoter Score and How Far to Trust It
Type in how many people scored 9–10, 7–8 and 0–6, or paste the raw answers from your survey export. You get the eNPS, and the range the true score for your whole workforce probably sits in — the number that tells you whether a five-point move since last quarter means anything.
Your survey answers
Optional. When many of your people answered, the margin of error shrinks (finite population correction). Leave 0 if unknown.
95% is the usual choice. Higher confidence gives a wider range.
38.0% promoters − 26.0% detractors, from 150 responses.
Promoters 57 · Passives 54 (36.0%) · Detractors 39
Margin of error ±7.7 points (adjusted Wald, corrected for 62.5% of 240 employees answering). The whole range is above zero: more of your people would recommend you than not.
+4.2 to +19.8
150 of 240
How the margin of error shrinks with more responses (same answer mix)
| Responses | Margin of error (95%) | An eNPS of +12.0 could really be |
|---|---|---|
| 25 | ±27.5 | about −16 to +40 |
| 50 | ±18.9 | about −7 to +31 |
| 100 | ±11.7 | about 0 to +24 |
| 150 (yours) | ±7.7 | about +4 to +20 |
| 200 | ±4.4 | about +8 to +16 |
Keeps your shares of promoters, passives and detractors and your headcount; only the number of responses changes. Ranges here are centred on your eNPS for readability, so they can differ by a point from the adjusted range above.
Worked example: 150 answers from a 240-person company
The calculator opens on a company of 240 employees that sent its pulse survey to everyone and got 150 answers: 57 promoters, 54 passives and 39 detractors. Here is every figure it shows, worked by hand.
| Step | Arithmetic | Result |
|---|---|---|
| Shares | 57 ÷ 150, 54 ÷ 150, 39 ÷ 150 | 38.0% promoters, 36.0% passives, 26.0% detractors |
| eNPS | 38.0 − 26.0 | +12.0 |
| Adjusted counts | add ¾ to promoters and detractors, 3 to the total | 57.75 and 39.75 of 153 |
| Adjusted shares and centre | 57.75 ÷ 153 − 39.75 ÷ 153 | 0.3775 − 0.2598 = +11.8 |
| Variance | 0.3775 + 0.2598 − 0.1176² | 0.6234 |
| Standard error | √(0.6235 ÷ 153) | 0.0638 |
| Finite population correction | √((240 − 150) ÷ (240 − 1)) | 0.614 |
| Margin of error, 95% | 1.96 × 0.0638 × 0.614 × 100 | ±7.7 points |
| Range | 11.8 − 7.7 to 11.8 + 7.7 | +4.1 to +19.4 |
The plain Wald version of the same sum, centred on +12.0 without the adjustment, gives ±7.8 and +4.2 to +19.8; the two differ little at 150 answers and a lot at 20. The response rate is 62.5%.
The correction is doing real work here. Leave the headcount at 0, as you would for a sample drawn from a huge customer base, and the same 150 answers give ±12.5 points and a range of −0.7 to +24.3 — which crosses zero, so you could not even say promoters outnumber detractors. Because 62.5% of the company answered, there is much less left unknown, and the range tightens to stay above zero.
Checked against a published example
Jim Lewis and Jeff Sauro of MeasuringU work the adjusted method through for a product survey with 15 promoters, 13 passives and 8 detractors (an NPS of +19.4) at 90% confidence and report a range of −2% to 38%. Enter those counts, leave the headcount at 0 and choose 90%: the calculator shows −2.4 to +38.3. The method itself comes from Brendan Rocks' 2016 paper on interval estimation for the NPS, which tested the options by simulation and concluded that the plain Wald and Goodman intervals “perform poorly; their use should be avoided.” Its best closed-form choice, the adjusted Wald labelled AW(3, T), is what the main range uses.
How to use the calculator
- Enter counts or paste scores. Most survey tools export the 0–10 answers as a column; paste it as it is. Blank cells, text and numbers outside 0–10 are skipped and counted, so you can see if anything was lost.
- Enter how many employees were invited. This is the population the score describes. If you surveyed one department, enter that department's headcount, not the company's.
- Choose a confidence level. 95% is the convention. 90% gives a narrower range you will be wrong about more often; 99% a wider one.
- Read the range, not just the score. If last quarter's eNPS sits inside this quarter's range, the honest summary is “no clear change.”
- Use the table under the calculator before the next survey: it shows how many answers you need for the precision you want.
Nothing you type or paste leaves your browser, so it is safe to paste a real export.
How many responses do you need?
Precision is bought with answers, and for a company of a given size the correction makes the last answers the most valuable. With the example's mix of answers and 240 employees, 100 responses give about ±11.7 points (an eNPS of +12 could be anywhere from 0 to +24), 150 give ±7.7, and 200 give ±4.4 (about +8 to +16). The same 100 answers from a company too big for the correction to matter would carry ±15.2. The table under the calculator redoes this for your own mix and headcount.
Why there is no “good eNPS” benchmark here
You will find many published eNPS benchmarks, and nearly all come from survey vendors summarising their own customers, with no public sample or method. We could not trace one to a source that would meet the standard of every other number on this page, so we left benchmarks out. The most useful comparison is your own previous survey, read with both ranges. Bain, which created the Net Promoter System, describes eNPS as “an emerging science.”
Common eNPS mistakes
- Averaging the 0–10 scores. An average of 7.4 is not an eNPS. Classify every answer first, then subtract the percentages.
- Dropping passives from the total. 57 promoters and 39 detractors out of 96 would give +18.8 instead of +12.0. Passives belong in the denominator.
- Reading small teams too closely. With 20 answers the margin of error is often ±25 points or more. Report small teams as part of a larger group, which also protects anonymity.
- Comparing scores from different questions. Changing the wording or the scale breaks the trend line; keep the question exactly the same between surveys.
- Treating a change as a result. A rise from +8 to +12 after a new recognition program is good news only if the ranges barely overlap. Pair eNPS with harder measures such as voluntary turnover.
Frequently asked questions
How do you calculate eNPS?
Ask employees how likely they are to recommend the company as a place to work on a 0–10 scale. Count 9s and 10s as promoters and 0–6 as detractors, then subtract the percentage of detractors from the percentage of promoters. 7s and 8s count only in the total.
What is the margin of error of an eNPS?
It depends on the number of answers and on how they split. At 95% confidence the plain formula is 1.96 × √((ppro + pdet − (ppro − pdet)²) ÷ n) on the 0–1 scale, multiplied by 100 for points. The calculator uses an adjusted version that holds up better with small samples.
Can eNPS be negative?
Yes. Any time detractors outnumber promoters the score is below zero; it can go as low as −100, where everyone answered 0–6.
Why does the number of employees change the margin of error?
The usual formulas assume a tiny sample from a huge population. In an employee survey a large share of the company often answers, so less is unknown. The finite population correction, √((N − n) ÷ (N − 1)), shrinks the margin to reflect that; if everyone answers, the margin is zero.
How often should you measure eNPS?
Often enough to see change, rarely enough that people still answer. Use the sample-size table to check that each survey will get enough responses for the difference you hope to detect.
Does this calculator store my survey data?
No. Everything is calculated in your browser; nothing you type or paste is sent or saved.
About the author
Maciej has worked in the digital gift card and rewards industry for over 5 years. For this tool he read Bain's Net Promoter System pages on scoring and on eNPS, Brendan Rocks' paper on NPS interval estimation, MeasuringU's articles on NPS confidence intervals and the finite population correction, and the NIST normal-distribution table on November 1, 2026; implemented the adjusted Wald interval, reproduced MeasuringU's published −2% to 38% example with it, and checked every figure in the worked example against the calculator's output.
- Bain & Company, Net Promoter System, “Measuring Your Net Promoter Score” — 0–10 scale; promoters 9–10, passives 7–8, detractors 0–6; % promoters minus % detractors: netpromotersystem.com
- Bain & Company, Net Promoter System, “Employee Net Promoter Score (eNPS)” — the “place to work” question; “an emerging science”: netpromotersystem.com
- Brendan Rocks, “Interval Estimation for the ‘Net Promoter Score’”, arXiv:1601.07235 (January 27, 2016) — NPS variance, Wald interval, adjusted Wald AW(3, T) recommendation: arxiv.org/abs/1601.07235
- Jim Lewis and Jeff Sauro, MeasuringU, “Confidence Intervals for Net Promoter Scores” (January 5, 2021) — adjusted Wald steps; 15/13/8 example at 90%: measuringu.com
- Jeff Sauro and Jim Lewis, MeasuringU, “Finite Population Correction” (January 3, 2023) — √((N − n)/(N − 1)), noticeable from about 10% sampled: measuringu.com
- NIST/SEMATECH e-Handbook of Statistical Methods, 1.3.6.7.1, Cumulative Distribution Function of the Standard Normal Distribution — z = 1.645, 1.960, 2.576: itl.nist.gov