Zurück
eNPS Survey: Formula, Questions and Benchmarks

Matthieu Saussaye

An eNPS survey takes one question to ask and a lot more thought to read correctly.
The employee Net Promoter Score asks people how likely they are to recommend their employer as a place to work, on a 0 to 10 scale. The number is easy to track, but on its own it tells you nothing about what to fix. The useful part is the follow-up question and the analysis of the answers. SmartInterview is the fastest way to run that kind of study: the AI drafts the survey, asks a follow-up when a comment is vague, and codes every comment into themes.
The Short Answer
The question: "On a scale of 0 to 10, how likely are you to recommend [company] as a place to work to a friend or acquaintance?"
The formula: eNPS = % promoters (9 to 10) minus % detractors (0 to 6). Passives (7 to 8) count in the total but not in either group. The score runs from -100 to +100.
Always add an open follow-up ("What is the main reason for your score?"). The score shows the direction; the comments show the cause.
Benchmarks are rough. Above 0 means more promoters than detractors. Vendor rules of thumb call +10 to +30 good and +50 excellent, but industry, country and company size move the number a lot. Your own trend is the best benchmark.
Run it quarterly or twice a year, keep it short, guarantee anonymity, and never report on groups smaller than five people.
eNPS Survey at a Glance
Element | What to do | Example |
|---|---|---|
Core question | 0 to 10 scale, labelled ends, same wording every wave | "How likely are you to recommend Acme as a place to work?" 0 = Not at all likely, 10 = Extremely likely |
Groups | Promoters 9 to 10, passives 7 to 8, detractors 0 to 6 | In a team of 20: 8 promoters, 7 passives, 5 detractors |
Formula | % promoters minus % detractors | 40% minus 25% = eNPS of +15 |
Follow-up | One open question, then a probe if the answer is vague | "What is the main reason for your score?" |
Driver questions | 3 to 6 short Likert items on known drivers | "My manager gives me useful feedback." Strongly disagree to Strongly agree |
Frequency | Quarterly or twice a year | Same month each cycle to avoid bonus or review effects |
Minimum sample to report | No breakdown under 5 respondents | Merge a team of 4 into its department |
Precision | Report a margin of error, not just the score | 120 answers: +15, plus or minus about 14 points |
The core question is a classic 11-point scale; for the customer version and its history, see our guide to the Net Promoter Score, and use the NPS calculator to check your numbers. The formula is identical for employees.
How the Employee Net Promoter Score Works
Where eNPS comes from
The Net Promoter Score was introduced by Fred Reichheld of Bain & Company in a 2003 Harvard Business Review article, "The One Number You Need to Grow". It was built for customers. Bain and others later applied the same question to employees, asking about the company as a place to work instead of its products. The scoring rules did not change, which is why eNPS is easy to explain to leadership and easy to compare over time.
The formula, step by step
Each employee gives a score from 0 to 10. You sort the answers into three groups, then subtract:
eNPS = (number of promoters / total respondents x 100) - (number of detractors / total respondents x 100)
Passives disappear from the formula but not from the denominator. That is the most common calculation mistake: if you divide by promoters plus detractors only, you inflate the score. The result is a whole number between -100 (everyone a detractor) and +100 (everyone a promoter). It is not a percentage, so write "+15", not "15%".
Worked example: a 150-person company
A software company with 150 employees runs its first eNPS survey and gets 120 answers (an 80% response rate).
Group | Scores | Respondents | Share |
|---|---|---|---|
Promoters | 9 to 10 | 48 | 40% |
Passives | 7 to 8 | 42 | 35% |
Detractors | 0 to 6 | 30 | 25% |
Total | 120 | 100% |
eNPS = 40 - 25 = +15.
How precise is that +15?
Every eNPS is an estimate, and small companies get noisy scores. Treat each answer as +1 (promoter), 0 (passive) or -1 (detractor). The standard error of the score, in points, is:
SE = 100 x √[(p + d - (p - d)²) / n], where p and d are the promoter and detractor shares and n the number of respondents.
Here: (0.40 + 0.25 - 0.15²) / 120 = 0.6275 / 120 = 0.00523. The square root is 0.0723, so SE = 7.2 points and the 95% margin of error is about 1.96 x 7.2 = plus or minus 14 points. The true score could be anywhere from about +1 to +29.
Because 120 out of 150 people answered, you can apply the finite population correction, √[(N - n) / (N - 1)] = √(30 / 149) = 0.45, which narrows the margin to about plus or minus 6 points. Either way, the lesson holds: a move from +15 to +20 next quarter is not a real change. Look for shifts larger than the margin, and for consistent trends over several waves. Our article on sampling methods explains why non-response matters as much as sample size.
eNPS vs engagement surveys
An engagement survey measures many dimensions (manager, growth, pay, workload, recognition) with 30 to 60 items, usually once a year. eNPS is one loyalty indicator you can ask often. They do different jobs, much like the split between qualitative and quantitative research: eNPS tells you whether sentiment is moving; the engagement survey, or a short set of driver questions, tells you where. A good setup uses eNPS as the headline in every pulse and rotates a few Likert-scale driver items around it. To find which drivers move the score, run a simple correlation analysis between each driver item and the 0 to 10 answer, keeping in mind that correlation is not causation.
Ready-to-Use eNPS Survey Questions
A complete eNPS pulse takes under five minutes. Here is a template you can copy, in the order respondents should see it. For general wording rules, see our survey questions guide.
1. The eNPS question
"On a scale of 0 to 10, how likely are you to recommend [company] as a place to work to a friend or acquaintance?" Answer options: 0 (Not at all likely) to 10 (Extremely likely). Keep this wording identical every wave, or your trend breaks.
2. The open follow-up
"What is the main reason for your score?" Free text. This is the most important question in the survey. Phrase it the same way for everyone; tailored versions ("What would make it a 10?" only for passives) make comparisons harder. More on writing these in our guide to open-ended questions.
3. The improvement question
"If you could change one thing about working here, what would it be?" Free text. It produces concrete, actionable suggestions even from promoters.
4. Driver statements
Five-point agreement scale (Strongly disagree, Disagree, Neither, Agree, Strongly agree):
"I have the tools and resources I need to do my job well."
"My manager gives me regular, useful feedback."
"I see a path to grow my skills or career here."
"My workload is manageable most weeks."
"I am fairly paid for the work I do."
"I understand how my work contributes to the company's goals."
5. A stay question
"Do you see yourself working here in 12 months?" Yes / No / Not sure. A short yes or no question with a "not sure" option is a useful early signal on retention, especially when you cross it with the eNPS groups.
6. Minimal demographics
Department (single choice) and tenure band (under 1 year, 1 to 3 years, 3 to 5 years, over 5 years). Use single-select questions with broad categories. Every extra field makes small groups identifiable, so ask only what you will actually use in the analysis.
7. Running it in SmartInterview
In SmartInterview, you type the objective, for example "quarterly eNPS pulse for 150 employees, find the top reasons for detractor scores", and the AI drafts the full questionnaire with a quality score before launch. You share one reusable link with your team. When someone answers the follow-up with "management" or "it's fine", the AI asks one more question to get the specific reason, and people can answer by text or voice, in any of 99+ languages. In our studies, AI follow-ups deliver up to 4x longer answers and 2.2x more topics. Every comment is then coded into themes linked to the original verbatims, with sentiment, and you can cross-tab themes by eNPS group, department or tenure, or export to Excel or SPSS. The platform is built and hosted in Switzerland and GDPR compliant, which matters when the data is about your own staff. Plans are on the pricing page.
Where I draw the line
I would not run an eNPS survey without the open follow-up, and I would not present the score without the margin of error. A bare number invites the wrong conversations: a leadership team celebrating +5 points that are pure noise, or a manager blamed for a drop that comes from four people in a team of nine. The score is a thermometer. Nobody fixes a fever by staring at a thermometer.
I also draw the line at tying eNPS to manager bonuses. The moment a score has money attached, people start asking their team for 9s, and the survey stops measuring anything. Use it to find problems and check that fixes worked, not to rank people.
Matthieu Saussaye, Founder of SmartInterview.
eNPS Survey Checklist
Write down the decision first. Retention in one department? Effect of a reorganisation? The goal decides which driver items you add.
Fix the core wording and scale and do not change them between waves.
Add one open follow-up and one improvement question. Without them, the score cannot be acted on.
Keep it under five minutes. Ten questions at most for a pulse.
Set an anonymity threshold (no results for groups under five) and say so in the invitation.
Choose a fixed rhythm: quarterly or twice a year, away from bonus and review periods.
Leave the survey open one to two weeks and send one reminder.
Report the score, the group shares, the response rate and the margin of error, not the score alone.
Read the comments by group. Detractor themes show what to fix, promoter themes show what to protect.
Share results and two or three actions within a month. Then measure again.
Common mistakes
Dividing by promoters plus detractors only. Passives belong in the denominator.
Reading small changes as trends. With 50 respondents, a swing of 20 points can be noise.
Comparing with a customer NPS or another industry's eNPS. Different questions, different populations.
Promising anonymity you cannot deliver. Department plus tenure plus gender in a 40-person company identifies people. Fewer demographics, higher thresholds. See survey bias for how fear of exposure distorts answers.
Surveying too often. Monthly eNPS with no visible action trains people to ignore it and lowers response rates.
Leaving comments unread. Hundreds of verbatims with no coding end up summarised by gut feeling. A tool from our list of survey analysis tools saves days.
Only acting on detractors. Passives are the largest group in many companies, and the easiest to move.
Frequently Asked Questions
What is a good eNPS score?
Any score above 0 means you have more promoters than detractors. Rules of thumb published by HR software vendors often treat +10 to +30 as good and +50 or more as excellent, but scores vary a lot by industry, country and company size. Your own trend over several waves is a more reliable benchmark than any external figure.
How do you calculate eNPS?
Ask the 0 to 10 recommendation question, then take the percentage of promoters (9 to 10) and subtract the percentage of detractors (0 to 6). Passives (7 to 8) are included in the total number of respondents. For example, 40% promoters and 25% detractors give an eNPS of +15.
What questions should an eNPS survey include?
The core 0 to 10 recommendation question, an open "What is the main reason for your score?", an improvement question, and three to six driver statements on topics like management, growth, workload and pay. Add only the demographics you need to break down the results.
How often should you run an eNPS survey?
Quarterly or twice a year works for most companies. That is often enough to see the effect of changes and rare enough for people to notice that something happened between waves. Avoid monthly surveys unless you can act on results every month.
Should an eNPS survey be anonymous?
Yes. People give more honest scores and comments when they are confident they cannot be identified. Use a minimum group size of five for any breakdown, limit demographic questions, and explain in the invitation who will see what.
What is the difference between eNPS and an engagement survey?
eNPS is a single loyalty question you can ask often to track sentiment. An engagement survey covers many dimensions with dozens of items, usually once a year, to explain why sentiment is where it is. Many companies run eNPS pulses between annual engagement surveys.
What is a good response rate for an eNPS survey?
Internal surveys usually get far higher response rates than customer surveys, and many HR teams aim for 70% or more. Low participation is itself a signal: it often means people doubt the anonymity or think nothing will change.
Want the reasons behind your score, not just the number? SmartInterview is free for your first 50 respondents.


