Net Promoter Score (NPS): how to calculate and read it

How to calculate the NPS, how precise it is for your sample and what a good score is. With a free NPS calculator, a worked example and templates.

The Net Promoter Score (NPS) is based on a single question: how likely is it that someone would recommend your company, brand or product to a friend, relative or colleague, on a scale from 0 to 10? Anyone who chooses 9 or 10 counts as a promoter, anyone who chooses 0 to 6 as a detractor. The NPS is the share of promoters minus the share of detractors, so it lies between −100 and +100 (Reichheld 2003, cited in Marx 2016, slide 77).

How much such a score tells you depends above all on how many answers are behind it. That's what this article is about, just as much as the calculation itself.

NPS calculator

I explain further down how the calculator gets from the number of answers per scale point to the NPS and its interval.

The example file shows how to arrange your data: Excel, CSV. It contains the 200 answers from the online store example below and also calculates the NPS with its 95% confidence interval directly in Excel.

The calculator runs in your browser and doesn't send any data.

The NPS question and the scale

The question is: "How likely is it that you would recommend company/brand/product X to a friend, relative or colleague?" The answer scale runs from 0 to 10, so it has eleven points (Marx 2016, slide 77). The answers are then sorted into three groups:

Take a look at how wide the groups are. Seven scale points belong to the detractors, only two to the promoters. In everyday life a 6 sounds like "pretty decent". In the NPS it counts as criticism. And an 8 doesn't count either way.

Fred Reichheld introduced the metric in the Harvard Business Review in 2003. His core claim: how many customers are willing to recommend a company is related to that company's growth (Reichheld 2003).

What if I use a shorter scale?

Some teams prefer a five- or seven-point scale, for example because the other questions in the questionnaire use that format too. You can still calculate a score, but it's no longer a classic NPS. I explain what to watch out for further down under NPS on a different scale.

How to calculate the NPS: formula and example

The formula is short:

Because the NPS is the difference between two percentages, I write it without a percent sign, so "NPS 26" and not "26%". I express differences between two scores in points.

A very small example

Let's look at ten guests in a café who answer the NPS question. Four of them choose a 9 or 10, three a 7 or 8 and three a number between 0 and 6.

• Promoters: 4 of 10 = 40%
• Detractors: 3 of 10 = 30%
• NPS = 40 − 30 = 10

The three passives don't show up in the last line. But they're in the denominator, because both shares refer to all ten guests.

Two extreme cases help you understand it. If all ten were promoters, the NPS would be 100 − 0 = +100. If all were detractors, it would be 0 − 100 = −100. It can't go higher or lower than that.

Worked example: 200 answers

Now a more realistic example. An online store asks 200 customers after delivery. This is how the answers are distributed (made-up numbers):

Step 1, count the groups:

• Detractors (0 to 6): 2 + 1 + 3 + 4 + 6 + 10 + 18 = 44
• Passives (7 and 8): 28 + 32 = 60
• Promoters (9 and 10): 46 + 50 = 96

As a check: 44 + 60 + 96 = 200.

Step 2, the shares of all 200 answers:

• Promoters: 96 / 200 = 48%
• Detractors: 44 / 200 = 22%

Step 3, the difference: NPS = 48 − 22 = 26.

A common mistake is to leave out the passives and calculate with only 140 answers. You'd then get 96 / 140 ≈ 68.6% promoters and 44 / 140 ≈ 31.4% detractors, so an NPS of ≈ 37. That's about eleven points too high, just because the denominator is wrong.

By the way, in QUESTIONSTAR the NPS is calculated automatically as soon as responses come in. You see the distribution in one of six chart types and can export the report as Word, Excel, PowerPoint or PDF, and the raw data as CSV, XLSX or SPSS. I describe what that looks like in detail under NPS, CSAT and CES in QUESTIONSTAR.

NPS on a different scale

The classic NPS is precisely defined: a scale from 0 to 10, detractors 0 to 6, passives 7 and 8, promoters 9 and 10 (Marx 2016, slide 77). Only a score measured this way can be compared with the scores of other companies or studies, and even then only with caution (see What is a good NPS?).

But you can calculate the difference on any scale. You take the share of the top group and subtract the share of the bottom group. The result is a net score. On a five-point scale it's a "net score on a 5-point scale", not an NPS in the strict sense. The report should label it that way too, so that nobody compares it with a classic NPS.

Where do the cutoffs go?

For a five-point scale from 1 to 5, there are two common approaches:

1. Only the top point counts as a promoter. Promoters are the 5s, the 4 is passive, detractors are 1 to 3. This is how Jeff Sauro analyzed the five-point scale in a comparison study (Sauro 2017). The idea behind it: in the classic NPS, too, only the top two of eleven points count.
2. The cutoffs follow the meaning of the scale points. Anyone who ticks "likely" or "definitely", so 4 or 5, is a promoter. The middle, the 3, is passive. Points 1 and 2 are detractors.

Both approaches can be justified. What matters is something else: set the cutoffs once and don't change them between two survey waves. Otherwise you're comparing two different metrics.

Example: the same answers, two results

Suppose 200 customers answered the recommendation question on a five-point scale:

Approach 1, only the 5 counts as a promoter:

• Promoters: 60 / 200 = 30%
• Detractors, points 1 to 3: (10 + 16 + 44) / 200 = 70 / 200 = 35%
• Net score: 30 − 35 = −5

Approach 2, cutoffs by the meaning of the scale points:

• Promoters, points 4 and 5: (70 + 60) / 200 = 130 / 200 = 65%
• Detractors, points 1 and 2: (10 + 16) / 200 = 26 / 200 = 13%
• Net score: 65 − 13 = 52

The same 200 answers therefore produce either −5 or 52, depending on the cutoffs. The 57-point difference comes from the cutoffs alone, not from the customers. That's why the cutoffs belong in every report, right next to the score.

Don't convert old data

If you used to ask with a five-point scale, those answers can't be converted into a "real" NPS after the fact. A 4 on the five-point scale doesn't correspond to any particular point between 0 and 10. So don't compare such scores with the NPS of other companies measured on the 0 to 10 scale either.

A study by Sauro (2017) shows how large the difference can get. The same respondents answered the recommendation question on the 0 to 10 scale, on a ten-point scale and on a five-point scale, with the five-point scale analyzed as in approach 1. On average, the net scores from the five-point scale were about 4 points away from the classic NPS, 8 points in the largest case, and there were clearly fewer promoters on the five-point scale. His advice: if you already have data on a different scale, stick with it. This is a single study with respondents from the US and one particular choice of cutoffs. For your data, the difference may be smaller or larger.

A three-point scale with "yes", "maybe" and "no" deviates even more. In three studies its net scores were 17 to 67 points away from the NPS on the 0 to 10 scale (Sauro 2019).

For surveys in several countries, you also need to check whether the highest rating reflects the same willingness to recommend in each country. This is a question of scale equivalence (Malhotra and Birks 2007, p. 759).

In QUESTIONSTAR and in the calculator

In QUESTIONSTAR you set the start and end of the scale freely and decide for each question which points count as detractors, passives and promoters. QUESTIONSTAR then calculates the net score automatically. You can also set your own scale and cutoffs in the calculator above under "Adjust scale and thresholds".

How precise is your NPS?

The online store measured an NPS of 26. Had it asked a different sample of 200 customers on another day, the result might have been 22 or 31. The confidence interval shows how much a result can vary purely because of random sampling. The calculator above adds exactly this interval, and the comparison of two measurements, to the QUESTIONSTAR report.

Step by step: the 95% interval

The trick is to write each answer as a number: a promoter counts +1, a passive 0, a detractor −1. The mean of these numbers over all 200 answers is (96 − 44) / 200 = 0.26. Multiplied by 100, that's exactly the NPS. So you can work with it like any other mean.

First we need the spread per answer, the variance. For the NPS there's a short formula for it (Rocks 2016, section 1.4):

Here p_P is the share of promoters and p_D the share of detractors, both as decimals, and s² is the variance.

In the example, with the shares as decimals: 0.48 + 0.22 − 0.26² = 0.70 − 0.0676 = 0.6324.

This gives the standard error, the typical fluctuation of the NPS with 200 answers:

In the example: SE = 100 · √(0.6324 / 200) ≈ 5.6 points.

For a 95% interval, you take the standard error times 1.96:

In the example: 1.96 · 5.6 ≈ 11 points. The interval therefore runs from 26 − 11 ≈ 15 to 26 + 11 ≈ 37.

For the measured NPS of 26, the 95% confidence interval is approximately 15 to 37. The 95% describes the procedure. If the store repeatedly drew new random samples of 200 customers and calculated an interval for each one, about 95 out of 100 of these intervals would contain the true value. Nobody knows whether your own interval is one of them. And the interval only captures this chance variation. If mainly the very satisfied or the very annoyed customers respond, it won't help you.

This procedure is called the Wald interval, and it's also how the calculator above works. It's an approximation. Rocks (2016, sections 2.1.1 and 3.2) compares it with methods that perform better. In his comparison, the average coverage of the Wald interval rises with the number of cases, but even at 300 answers it's still below the intended 95%. That's why the calculator flags a small sample when there are fewer than 30 answers. The approximation can be especially unreliable then. A narrow interval doesn't automatically mean high precision. If all answers fall into the same group, the Wald formula gives a standard error of zero. That doesn't mean certainty about the true NPS. Such an interval isn't meaningful.

Comparing two measurements

Often the single score matters less than the question: have we improved? Let's say a small specialist retailer asks 50 customers in spring and another 50 in fall (made-up numbers). For this example we assume two random samples drawn independently of each other. If the same people are surveyed again, the comparison needs an analysis for paired answers.

The NPS dropped by 6 points. Is that a warning sign? Let's do the math.

To find out, you divide the difference by its standard error:

The standard error of the difference is √(11.0² + 11.5²) ≈ 16 points. So the 6-point difference is only a little more than a third of that: 6 / 16 ≈ 0.38. This value is called z. In the two-sided test used here, an absolute z of about 1.96 or more counts as statistically notable at the 5% level. The corresponding p-value is ≈ 0.71. A difference of this size between two samples of 50 answers each can therefore easily arise by chance, even if nothing has actually changed.

You can also see it in the intervals: they overlap almost completely. With these distributions, a six-point difference can't be statistically established with 50 answers per measurement. That's not a weakness of the retailer. It's math. The calculator above does exactly this comparison when you enter a second measurement under "Compare two values".

How many answers do you need?

You can rearrange the formula to calculate the required sample size. If you want your NPS to be accurate to ±10 points (95%), you need:

Here E is the desired precision as a decimal, so 0.10 for ±10 points.

With the variance from our online store (0.6324), that's 3.8416 · 0.6324 / 0.01 ≈ 242.9, rounded up to 243 answers. If you don't know the distribution yet, assume the worst case, a variance of 1. Then it's ≈ 384.2, rounded up to 385 answers. This worst case occurs when half are promoters and the other half detractors.

For ±5 points you need four times as many answers, ≈ 972 in the example and ≈ 1,537 in the worst case. You can read more about the logic behind this in the article Calculating sample size.

The follow-up question: why?

An NPS of 26 tells you what the mood is. It doesn't tell you what's behind it. That's why the NPS question is usually followed by an open question, for example: "What is the most important reason for your rating?"

Open questions give rich answers in the customers' own words. Analyzing them takes time because someone has to read and categorize the responses (Marx 2026, slide 81). My advice: read the answers separately for detractors, passives and promoters, and note the topics that come up again and again, such as delivery time, packaging or customer service. Then count how often each topic is mentioned in each group. You'll often find that detractors write about completely different things than promoters.

By the way, in QUESTIONSTAR you can add a follow-up question to the NPS question and get an email notification when someone gives a low rating. That way someone from your team can get in touch promptly. You can also send the answers to your CRM via a webhook.

What is a good NPS?

This is the most common question about the NPS, and the answer is an uncomfortable one: there's no universal threshold. How high an NPS turns out depends on the industry, on the exact wording of the question and on the channel you use to ask. A score from a survey right after a hotline call therefore can't simply be compared with a score from an email survey of all customers.

That's why QUESTIONSTAR doesn't provide industry benchmarks. Reichheld himself also advises caution. Later, together with Markey, he writes that the recommendation question is the best single question for a range of industries, but not for all of them. Every company has to check for its own business whether the answers are related to customers' later behavior (Reichheld and Markey 2011, pp. 50-51, cited in Marx 2016, slide 78).

My advice: compare yourself with yourself. Always ask the same way over time, with the same wording, through the same channel and at the same point in the customer relationship. Look at the interval for every change, as shown in Comparing two measurements. And read the reasons people give. What changes there is often more revealing than the number.

A positive NPS means that more promoters than detractors answered in your sample, a negative one the opposite. The calculator shows you this statement ("More promoters than detractors." or "More detractors than promoters.") together with the "95% confidence interval" on the scale from −100 to +100. If the interval contains 0, the sign of the NPS in the population isn't statistically established with the approximation described, and the calculator says "Sign not established." The calculator deliberately gives no rating such as "good" or "excellent".

Limitations and criticism

In my survey research course, the NPS appears in the chapter on reliability and validity, with two questions: is the scale reliable? Is the scale valid? (Marx 2016, slide 77). Neither can be taken for granted with a single question.

The criticism mainly targets Reichheld's core claim. Keiningham, Cooil, Andreassen and Aksoy (2007) examined longitudinal data from 21 companies in the Norwegian Customer Satisfaction Barometer. They couldn't confirm the claimed clear superiority of the NPS in predicting growth. In addition, they compared reconstructed NPS data from industries Reichheld had cited with the ACSI. In two of three industries, the coefficient of determination R² for the relationship between the ACSI and growth was higher than for the NPS (Keiningham et al. 2007, p. 45).

Then there's a practical warning. A report by Harvard Business Review Analytic Services (2019, p. 15) quotes the observation that many people responsible for customer experience are fixated on metrics such as NPS or CSAT instead of paying attention to the value for the business.

And finally, the calculation itself. Grouping answers into three groups loses information. Let's look at two extremes. A company where half the respondents are promoters and the other half detractors has an NPS of 0. A company where everyone gives a 7 or 8 does too. The two have very different customers. Similarly: if someone moves from 0 to 6, the NPS doesn't change. If someone moves from 8 to 9, it does.

So here's what the NPS doesn't tell you:

• why customers answer the way they do (that's what the follow-up question is for),
• who exactly the detractors are and what they have in common (that's what group comparisons are for),
• whether someone actually recommends you, because the question asks about likelihood, not behavior.

None of this argues against measuring the NPS. It argues against measuring it alone.

More than one question: the Secure Customer Index

Satisfied customers do not necessarily come back. Many don't buy again despite being satisfied (Reichheld 1997, cited in Braun and Müssigmann 2009, p. 140). That's why there are approaches that combine several questions into one picture. One of them is the Secure Customer Index (SCI). In my course I trace it back to D. Randall Brandt at Maritz Research (Brandt 1996, cited in Marx 2016, slide 70). Burke, Inc. describes it as its own method (Burke 2007, p. 2). From the sources available to me, I can't say for sure who developed it first. For the calculation it doesn't matter, because the rules are the same.

The SCI asks three questions, each on a scale from 1 to 5. The labels for recommendation and reuse are those of my English course (Marx 2016, slide 70). The five satisfaction points follow the current German version of the course (Marx 2026, slide 70):

Points 2 and 4 lie symmetrically in between: "somewhat dissatisfied" and "somewhat satisfied", or "will probably not ..." and "will probably ..." (Marx 2016, slide 70; Marx 2026, slide 70).

The four segments

Only someone who chooses the top point 5 on all three questions counts as a "Secure Customer" (Marx 2016, slide 70; Marx 2026, slide 70). Here the index is the share of these people among all respondents with valid answers to all three questions. Report cases with missing or invalid answers separately. The other respondents are assigned to three further segments. The rules come from Burke, Inc. (Baker-Prewitt 2007, slide 13):

• Secure: the best point on all three questions.
• Favorable: the best or second-best point on all three questions, but not Secure.
• At risk: the worst or second-worst point on at least one question.
• Indifferent (sometimes "Vulnerable", as with Brandt and Maritz): all other combinations.

The order in which you check matters. This is how I check:

1. Does the person have three 5s? Then they're Secure.
2. If not: are all three answers 4 or 5? Then Favorable.
3. If not: is at least one answer a 1 or 2? Then At risk.
4. Everyone left over is Indifferent.

Why At risk before Indifferent? A person with 5, 3 and 2 has a middle answer and at first glance could look indifferent. But the 2 settles it: this person is at risk.

Worked example: ten customers

Let's look at ten customers who answered all three questions (made-up answers, in the order satisfaction, recommendation, repurchase):

Result: 2 of 10 are Secure, so the Secure Customer Index is 20%. In addition, 30% are Favorable, 20% Indifferent and 30% At risk.

Look at person 8. She's somewhat dissatisfied, but says she would definitely recommend you and definitely continue to use you. Maybe she has no alternative. Had you asked only about recommendation, she'd look like an enthusiastic customer. These are exactly the cases the SCI makes visible.

Burke, Inc. later extended the index to five questions. The additions are whether the company has earned the customer's loyalty and whether the customer prefers it to all other providers. Burke also links the index from one period to the share of spending that customers give this company in the following period, the share of wallet (Baker-Prewitt 2007, slides 12 and 16; Marx 2016, slide 71).

How to calculate it with your data

You can calculate the SCI directly in the calculator above: first switch to "SCI", then choose "Paste three columns" or "Load file". It counts only people with valid answers to all three questions and shows the excluded cases separately. You'll find more on the index, its segments and how to read them in the article Secure Customer Index.

In QUESTIONSTAR, first use a compound filter to find the number of people who have a valid answer value from 1 to 5 on each of the three questions. This number is your base. Then use a second filter to count those who chose a 5 on all three questions. The Secure Customer Index is this count divided by the base you determined first, times 100. Report cases with missing or invalid answers separately.

There are two ways to get all four segments. You export the raw data as a CSV, Excel or SPSS file and assign the people to segments in Excel, SPSS or R using the four rules. Or you connect QUESTIONSTAR to an AI assistant such as ChatGPT or Claude via MCP and ask it, for example: "Consider only people with valid answers from 1 to 5 on all three questions 4, 5 and 6. Assign each of these people to a segment. Check in this order: Secure if all three answers are 5. Favorable if all three are 4 or 5. At risk if at least one answer is 1 or 2. Everyone else is Indifferent. Give the count and share per segment, based on these people with complete and valid answers, as well as their total number. Report the number of cases excluded because of missing or invalid answers separately." The assistant does the calculation, not QUESTIONSTAR. Check two or three people by hand.

Questions marked as personal data are not transferred to the assistant when the protection setting is on. Unmarked answers, especially free text, can still contain personal information. Before transferring, check which answers go to the AI provider and whether its contract terms suit your data.

NPS, CSAT, CES: which metric for what?

Besides the NPS, you'll come across two other three-letter metrics. They measure different things.

The Customer Satisfaction Score (CSAT) asks about satisfaction, often with a specific interaction, such as a purchase or a conversation with customer service. Usually it counts the share in the top two points, the so-called top-2-box. If only the highest ratings are counted, that's called top-box (Linke 2018, p. 105). More on this in the article Customer Satisfaction Score (CSAT).

The Customer Effort Score (CES) asks how easy it was for the customer to get their issue resolved. It goes back to Dixon, Freeman and Toman (2010). In Gartner's version, respondents rate their agreement with a statement on a scale from 1 to 7, and the share with 5 or higher is counted (Gartner 2018). More on this in the article Customer Effort Score (CES).

My advice: the NPS is the right choice when you want to know how customers feel about you overall. CSAT and CES are a better fit for a single touchpoint that you want to improve. You can switch the calculator above to all three.

eNPS: the NPS for employees

There's also a version of the NPS for employees, the Employee Net Promoter Score (eNPS). It asks how likely someone would be to recommend the company as an employer. The calculation is exactly the same as above: share 9-10 minus share 0-6. The calculator above works for it too.

Templates for your NPS survey

If you don't want to start from scratch, you'll find these templates with recommendation questions in QUESTIONSTAR:

• Net Promoter Score: standard with demographics
• Support-website rating: CSAT + Task Completion + recommendation (NPS-like)
• Hotline call rating: post-call survey

You copy a template into your account with a button inside the template once you've opened it. If you don't have an account yet, register for free first. Then check whether the follow-up question fits your occasion, and decide when you'll ask again. You'll find all templates on the topic under Templates.

Frequently asked questions about the Net Promoter Score

How do you calculate the NPS?

You count how many respondents chose 9 or 10 (promoters) and how many chose 0 to 6 (detractors). You divide both numbers by all answers, including the passives with 7 or 8. Then you subtract the share of detractors from the share of promoters. With 48% promoters and 22% detractors, that gives an NPS of 26.

What is a good NPS?

There's no universal threshold, because the score depends on the industry, the wording of the question and the survey channel. Compare your NPS with your own earlier measurements and pay attention to the interval. More on this above.

Why does the scale run from 0 to 10?

Because that's how the NPS is defined: the cutoffs 0 to 6, 7 to 8 and 9 to 10 refer to these eleven points (Marx 2016, slide 77). With a different scale and your own cutoffs you get a net score whose values aren't comparable with the classic NPS. More on this under NPS on a different scale.

How many responses do you need for the NPS?

For an approximate margin of error of ±10 points at a 95% confidence level, planning for independent random observations gives 385 answers in the worst case, and about 243 for a distribution like the one in our example. For the two distributions in the comparison example, the margin of error with 50 answers per measurement is more than ±20 points. Below 30 answers, interpret the Wald approximation with particular caution. You'll find the calculation above.

Can the NPS be negative?

Yes. If more detractors than promoters answer, the NPS is negative. The lowest possible score is −100, when all respondents choose 0 to 6.

What's the difference between NPS and CSAT?

The NPS asks about the willingness to recommend and subtracts detractors from promoters. CSAT asks about satisfaction and usually counts only the share in the top two points. That's why the NPS runs from −100 to +100 and CSAT from 0 to 100%. More in the article CSAT.

Sources

Baker-Prewitt, J. (2007). Keeping Customer Relationship Measurement Programs Relevant: Using Your Resources Wisely. Burke, Inc., presentation, North American Conference on Customer Management, October 2007. Slides 12, 13 and 16.

Brandt, D. R. (1996). Secure Customer Index. Maritz Research. Cited in Marx (2016), slide 70.

Braun, O. L. and Müssigmann, M. J. (eds.) (2009). Kundenbefragungen. Eine Sammlung von Fallstudien [Customer surveys: a collection of case studies]. Pabst. In German. p. 140.

Burke, Inc. (2007). Advances in Customer Loyalty Measurement. White paper, Pub. No. 0702. p. 2.

Dixon, M., Freeman, K. and Toman, N. (2010). Stop Trying to Delight Your Customers. Harvard Business Review, July-August 2010. hbr.org

Gartner (2018). What's Your Customer Effort Score? Article by Jordan Bryan, July 12, 2018. Archived original version

Harvard Business Review Analytic Services (2019). Beyond Big Data. Why Small Data Integration Is the Key to CXM Success. p. 15.

Keiningham, T. L., Cooil, B., Andreassen, T. W. and Aksoy, L. (2007). A Longitudinal Examination of Net Promoter and Firm Revenue Growth. Journal of Marketing, 71(3), 39-51. DOI 10.1509/jmkg.71.3.039

Linke, R. (2018). Mitarbeiterbefragungen optimieren. Von der Befragung zum wirksamen Management-Instrument [Optimizing employee surveys: from the survey to an effective management tool]. Wiesbaden: Springer Gabler. In German. p. 105.

Malhotra, N. K. and Birks, D. F. (2007). Marketing Research. An Applied Approach (3rd European ed.). Harlow: Pearson. p. 759.

Marx, P. (2016). Principles of Survey Research (slide deck). Slide 70 (Secure Customer Index), slide 71 (extended Secure Customer Index), slides 77-78 (Net Promoter Score).

Marx, P. (2026). Grundlagen der Umfrageforschung [Principles of survey research] (slide deck). In German. Slide 70 (Secure Customer Index, five-point scales), slide 81 (open and closed questions).

Reichheld, F. F. (1997). Der Loyalitäts-Effekt [The loyalty effect]. Frankfurt/Main: Campus. In German. Cited in Braun and Müssigmann (2009), p. 140.

Reichheld, F. F. (2003). The One Number You Need to Grow. Harvard Business Review, 81(12). hbr.org

Reichheld, F. and Markey, R. (2011). The Ultimate Question 2.0. Harvard Business Review Press. pp. 50-51. Cited in Marx (2016), slide 78.

Rocks, B. (2016). Interval Estimation for the "Net Promoter Score". The American Statistician, 70(4), 365-372. arXiv:1601.07235. Sections 1.4, 2.1.1 and 3.2.

Sauro, J. (2017). Changing the Net Promoter Scale: How Much Does It Matter? MeasuringU, November 29, 2017. measuringu.com

Sauro, J. (2019). Can You Use a 3-Point Instead of an 11-Point Scale for the NPS? MeasuringU, August 21, 2019. measuringu.com

Net Promoter and NPS are registered trademarks.

I hope the NPS is now more than just a number for you.

I wish you many promoters and revealing answers to the question of why!

Dr. Paul Marx

Author: Dr. Paul Marx
Dr. Paul Marx is the founder and Managing Director of QUESTIONSTAR and was a professor of marketing at the University of Siegen.