20 Min
Net Promoter Score (NPS) is a customer loyalty metric built on one question: how likely are you to recommend us to a friend or colleague, on a scale of 0 to 10? To calculate it, subtract the percentage of detractors (scores 0 to 6) from the percentage of promoters (scores 9 or 10). The result runs from -100 to +100. Any score above 0 is positive, above 30 is strong and above 50 is excellent.
Most companies measure the Net Promoter Score. Far fewer can say what it is telling them. The number lands in a monthly deck, someone notes that it moved two points, and the meeting continues. Nobody acts. That is a waste of the one metric customers answer honestly, because it asks something they genuinely have an opinion about.
This guide covers what the score is, how to calculate it with a worked example, what counts as good, which survey questions produce answers you can act on, and the specific changes that move the number. It also covers what the score cannot tell you. That part matters just as much.
Table of contents
- What is Net Promoter Score?
- How to calculate Net Promoter Score
- What counts as a strong score
- The survey questions to ask
- When to send it, and how often
- How to improve Net Promoter Score
- Employee Net Promoter Score
- The benefits, and the limits
- Where the score sits among your other metrics
- Frequently asked questions about Net Promoter Score
What is Net Promoter Score?
Net Promoter Score is a loyalty measure built on one question: how likely are you to recommend us to a friend or colleague? Customers answer on a scale from 0 to 10. Their answers sort into three groups. The score is the gap between the top group and the bottom one.
The metric came out of work by Frederick F. Reichheld, published in Harvard Business Review in December 2003 under the title “The One Number You Need to Grow”. His argument was that a single recommendation question predicted growth better than long satisfaction questionnaires did. Two decades on, it is the most widely reported customer metric in business. That ubiquity is both its strength and its weakness, because everyone has a score, so everyone compares, and most of those comparisons are meaningless.
Three groups come out of the 0 to 10 scale.
Promoters score 9 or 10. They come back, and they bring other people with them. Passives score 7 or 8. Satisfied but unattached, they will move for a better offer without much hesitation. Detractors score 0 to 6. They are unhappy, and a few of them say so in public where it costs you future customers.
How to calculate Net Promoter Score
The net promoter score formula is simple, and the one thing people get wrong is the passives.
Take the percentage of respondents who are promoters. Subtract the percentage who are detractors. Ignore the passives entirely. They sit in the denominator but never in the result, which is the single most common mistake in a hand-built spreadsheet version of this calculation. The score runs from -100, where every respondent is a detractor, to +100, where every one is a promoter. Write it as a whole number. Never as a percentage.
| Step | What you do | Worked example |
|---|---|---|
| 1 | Count responses by band | 500 responses: 260 promoters, 140 passives, 100 detractors |
| 2 | Convert promoters to a percentage | 260 / 500 = 52% |
| 3 | Convert detractors to a percentage | 100 / 500 = 20% |
| 4 | Subtract detractors from promoters | 52, 20 = 32 |
| 5 | Write it as a whole number | NPS is 32 |
A net promoter score calculator does the same arithmetic. The tool matters far less than the sample behind it. A clean, representative sample of a few hundred customers beats a large one skewed toward the people who always answer. A pair of rules protects the number from the two failure modes that make executive reporting worthless, which are a sample too small to mean anything and an audience that quietly changes between one quarter and the next. Report the response count beside the score, because a score of 40 from 30 responses is noise. And keep the audience consistent between periods, because a score that only surveys customers who contacted support is measuring something different from one that surveys everybody.
What counts as a strong score
Any score above zero means promoters outnumber detractors. A result over 30 counts as strong, and anything over 50 is excellent by the conventions most benchmark reports use. Those thresholds are conventions rather than laws, and they travel badly between industries, between survey designs and between countries, which is why a number quoted without its context is close to meaningless.
Net promoter score benchmarks are worth treating with suspicion for three reasons. Industries differ structurally, so a subscription software company and an airline are not on the same scale. Survey timing changes everything, because asking right after a resolved contact produces a different score than asking at renewal. And cultures score differently, since respondents in some markets rarely give a 10 at all.
Compare against yourself instead. Track your own score across consistent periods, split it by segment and by contact type, and watch the direction. A competitor’s published score tells you almost nothing about your business.
How to read net promoter scores by range
Net promoter scores can fall anywhere from -100 to +100. The table below shows how to read each range, using the same conventions as above. Treat it as a starting point rather than a benchmark for your industry.
| Score range | What it means | What to do next |
|---|---|---|
| -100 to -1 | Detractors outnumber promoters | Read every detractor comment and fix the most common cause first |
| 0 | Promoters and detractors are equal | Find what separates the customers who score high from those who score low |
| 1 to 30 | Positive: more promoters than detractors | Move passives up by fixing repeat and unresolved contacts |
| 31 to 50 | Strong | Protect what works and track the trend by segment |
| 51 to 100 | Excellent | Check the sample size and audience before you celebrate |
A negative result is calculated the same way. Say 200 customers respond: 50 promoters, 70 passives and 80 detractors. Promoters are 25% of responses and detractors are 40%, so the score is 25 minus 40, which is -15.
To see how the score connects to day-to-day service, compare it with your customer satisfaction rate. Then use these ways to improve satisfaction to move passives toward a 9 or 10.
Knowledge Management For a Higher CX Standard
The survey questions to ask
A net promoter score survey has two questions, and the second one carries all the value.
The first is the standard rating question, asking how likely the customer is to recommend you on a scale of 0 to 10, worded exactly as it was the previous quarter and the quarter before that. Keep the wording stable forever, because changing it breaks comparability with every previous period.
The second is open: what is the main reason for your score. This is where the metric stops being a number and starts being a to-do list. Without it you know that 20% of customers are detractors and nothing about why.
Some teams add a third, and the best version asks what one change would raise your score. It produces specific, actionable answers rather than general complaints. Resist anything more. Each extra question lowers the response rate, and a representative sample matters more than extra fields.
For a net promoter score survey template, four elements are enough: the rating question, the open follow-up, an optional single improvement question, and a clear statement of how long it takes. Send it in the channel the customer already used. Keep it to one screen on a phone.
When to send it, and how often
Timing decides what the score actually measures, so pick one of two models and stay with it.
Relationship surveys go to a sample of all customers on a fixed cycle, usually quarterly, and measure how people feel about the company. Transactional surveys go out after a specific interaction, such as a resolved support contact, and measure that experience. Both are legitimate. Mixing them in one reported number is not, because you can no longer tell whether a change came from the product, the price or the support queue.
Transactional surveys pair naturally with the operational numbers the support team already watches, so read the score beside average handle time for the same contact type rather than on its own. Whatever the cycle, cap how often any individual customer is asked. Once a quarter is a reasonable ceiling. Survey fatigue does more than lower response rates. It skews them, quietly and consistently, toward the people holding the strongest opinions in either direction. That is exactly the population you least want over-represented in a measure of loyalty, and the effect compounds every quarter you keep asking the same names.
How to improve Net Promoter Score
Scores move only when the experience behind them changes. So the useful work starts in the verbatim comments rather than in the number itself, and the method is unglamorous: read every comment from the period, group them by underlying cause, then count how many sit in each group. That ranked list is the roadmap. It is short.
Four causes come up again and again in support-driven detraction.
- The customer had to repeat themselves. For example, a transfer loses the context and the customer explains the problem twice. The fix is a shared record of the contact that travels with the transfer, so the second agent opens with the facts already in hand.
- Different answers from different agents. For example, two agents quote two refund windows and the customer catches the contradiction. The fix is a single owned article per policy, which is the core of any customer service knowledge base that agents actually trust.
- The first answer was wrong or incomplete. For example, the case closes and the customer contacts again four days later. The fix is guided steps for procedures with conditions, so the agent cannot skip the branch that matters.
- Self-service sent them in a circle. For example, the help center article exists but never surfaces for the words the customer used. Gartner surveyed 5,728 customers and found that self-service fully resolved only 14% of issues, with 43% of the failures caused by customers not finding relevant content. Every one of those customers arrives at the queue already annoyed, which is why self-service platforms have to be measured on resolution rather than on deflection.
Close the loop with detractors
Closing the loop matters as much as fixing the causes. Contact the detractors who left a comment, tell them specifically what changed as a result of what they wrote, and a meaningful share of them will revise their score without being asked to. Support groups that do this consistently see the number move before the underlying experience has fully caught up, because the follow-up is itself part of the experience.
Read the Full Case Study
Employee Net Promoter Score
Employee Net Promoter Score runs the same arithmetic on staff. The question becomes how likely you are to recommend this company as a place to work, and the bands and formula stay identical.
It is worth running alongside the customer score in a contact center, because the two move together more often than not. Agents who cannot find answers, apologize all day for problems they did not create and get measured on handle time tend to score their employer low. Their customers then score the company low for closely related reasons, usually on the same handful of unresolved issues. When both numbers fall in the same quarter, the cause usually sits in the tooling rather than in either group of people.
The benefits, and the limits
The benefits of net promoter score are real but narrow. One question gets far better response rates than a long survey. The single number travels across an organization without explanation. Tracked consistently, the trend is a genuine early signal. And the open follow-up produces qualitative insight that no operational metric delivers.
The limits are equally real. Knowing them is what keeps the metric honest, and none of the four below is a reason to stop measuring.
The score is slow, so it confirms what operational data showed weeks earlier. It is coarse, since a 6 and a 0 count the same. It is easy to game, because agents who ask for a 10 will get more of them and learn nothing. And it is lagging by construction, which is why it belongs beside first contact resolution, repeat contact rate and handle time by contact type rather than above them. Teams that run it as one input among several get value from it. Run it as the headline number instead, and you end up optimizing the survey rather than the service.
Where the score sits among your other metrics
No single number runs a support operation, and the score works best as the slowest dial on a panel of four.
Put first contact resolution next to it. Unresolved contacts create detractors faster than anything else does, and the link between the two is direct enough to show up within a single quarter. Then add repeat contact rate within seven days, which catches the answers that were wrong rather than merely slow.
Finish with a satisfaction measure taken immediately after the contact. The choice of NPS versus CSAT is less a rivalry than a question of timescale, because one asks about the relationship and the other about the moment.
Read together, the four tell a coherent story. Operational numbers move first, satisfaction follows within weeks, and the loyalty score confirms the change a quarter later. When the loyalty score moves and none of the others did, be suspicious of the sample before you celebrate.
Ready to Build Your Own Customer Service Knowledge Base?
Frequently asked questions about Net Promoter Score
Anything above zero means you have more promoters than detractors. Scores over 30 are usually considered strong and those over 50 excellent, but the thresholds shift by industry, by survey timing and by country. The comparison that matters is your own score over consistent periods with a consistent audience, rather than a competitor’s published figure.
Work out the percentage of respondents scoring 9 or 10, then subtract the percentage scoring 0 to 6. Passives scoring 7 or 8 are ignored in the subtraction but still count in the total. If 52% are promoters and 20% are detractors, the score is 32. Write it as a whole number from -100 to +100, never as a percentage.
Two are enough. Ask how likely the customer is to recommend you on a 0 to 10 scale, then ask openly what the main reason for that score is. The open question turns a number into a list of causes. A third question about the one change that would raise the score is worth adding, and more than that costs response rate.
For relationship surveys, quarterly to a rotating sample works for most companies. For transactional surveys, send after a specific interaction such as a resolved support contact. Cap any individual customer at roughly one survey a quarter, and never mix relationship and transactional results into one reported number, because the two measure different things.
Yes. A score below zero means detractors outnumber promoters. The scale runs from -100, where every respondent scores 0 to 6, to +100, where every respondent scores 9 or 10. For example, 25% promoters and 40% detractors give a score of -15. A negative score means the detractor comments are the place to start.
They are the three groups that answers on the 0 to 10 scale fall into. Promoters score 9 or 10 and are loyal enough to recommend you. Passives score 7 or 8 and are satisfied but open to a better offer. Detractors score 0 to 6 and are unhappy enough to warn other people away.
Frederick F. Reichheld introduced it in the Harvard Business Review article “The One Number You Need to Grow”, published in December 2003. He argued that one question about willingness to recommend predicted company growth better than long satisfaction questionnaires. That single recommendation question is still the core of every NPS survey.
NPS measures loyalty to the company over time, using one recommendation question on a 0 to 10 scale. CSAT measures satisfaction with a single interaction, usually right after it ends. Use CSAT to judge a specific contact and NPS to track the relationship. Tracking both shows whether good contacts are building lasting loyalty.
Word Document)
Excel File)
