Net Promoter Score (NPS) – Best practice

Net Promoter Score

The Net Promoter Score (NPS) is a simple but powerful tool to measure client satisfaction with one single question, an indication of the growth potential of your company or product. Read further for an overview on the use, application and pitfalls of NPS.

What is the Net Promoter Score?

The Net Promoter Score is a customer loyalty metric developed in 2003 by management consultant Fred Reichheld of Bain & Company in collaboration with the company Satmetrix. The objective was to determine a clear and easily interpretable customer satisfaction score which can be compared over time or between different industries.

The NPS assesses to what extent a respondent would recommend a certain company, product or service to his friends, relatives or colleagues. The idea is simple: if you like using a certain product or doing business with a particular company, you like to share this experience with others.

Specifically, the respondent is asked the following question:

How likely are you to recommend company/brand/product X
to a friend/colleague/relative?

This can be answered on an 11-point rating scale, ranging from 0 (not at all likely) to 10 (extremely likely).


Depending on the score that is given to the Net Promoter question, three categories of people can be distinguished:

  • Promoters = respondents giving a 9 or 10 score
  • Passives = respondents giving a 7 or 8 score
  • Detractors = respondents giving a 0 to 6 score


Net Promoter Score


The Net Promoter Score is calculated as the difference between the percentage of Promoters and Detractors. The NPS is not expressed as a percentage but as an absolute number lying between -100 and +100.

If you have for example 25% Promoters, 55% Passives and 20% Detractors, the NPS will be +5. A positive NPS (>0) is generally considered as good.

Don’t make the common mistake of placing a percent sign (%) behind your NPS score, it is not a percent.


NPS is used nowadays by many large companies as a customer feedback tool. It gives your organization an unambiguous number that is easy to understand for all employees and useful as input for managers to steer the company . According to many people the NPS also gives a good indication of growth potential and customer loyalty for a company or product.

You can track the evolution of the NPS over time, or compare it with a predetermined target. You can also benchmark different areas or products, or check where your company positions itself versus the industry average if this is available.

To give an indication: according to Reichheld the average American company scores less than +10 on the NPS, while the highest performing organizations are situated between +50 and +80. These values may however vary considerably from sector to sector and from culture to culture.

To understand the motives of Promoters and Detractors it is recommended to accompany the NPS question by one or more open questions that probe the underlying reasons behind the given score. This allows you to make the appropriate adjustments to increase the future NPS, either by boosting the percentage of Promoters, either by reducing the proportion of Passives and Detractors (or better yet, a combination of both).

NPS in the CheckMarket tool

To respond to the increasing popularity of the Net Promoter Score CheckMarket has decided to add the NPS question as standard question in the tool. If you want to use the NPS in your survey, the only thing you have to do is to select this question type. It is still possible to modify the formulation of the question or the naming of the endpoint value labels.

Once the responses start rolling in, you will see a bar chart in the reporting tool, showing the percentage of Detractors (red), Passives (orange) and Promoters (green). Under that is the detailed response distribution for the 11 possible scores.

Based on these percentages, the actual NPS is calculated. Your NPS is represented on an easy-to-read gauge (showing positive NPS in green, negative in red). Enabling you to see at a glance how well your product or business is performing.

Net Promoter Score - reporting


Some critical remarks

From a scientific perspective and in certain market research circles there is some skepticism about the NPS. Opponents of the NPS concept argue that there is insufficient scientific base for the outcome and that the model is too simple. They claim customer loyalty and satisfaction is not only about numbers and percentages, but also about causes, consequences and correlations.

It is also suggested that the NPS method cannot accurately measure customer behavior. Customers can claim they will recommend a company or product in a greater or lesser extent, but it is not proven they will actually do that in practice. Besides that, the recommendation of one customer is not always as valuable as that of another one.

It is also a fact that the NPS is more useful in markets with a lot of competition where potential buyers have a greater tendency to ask friends or acquaintances for advice before deciding about a purchase. Another limitation of NPS is that it only takes into account customers, while also a lot of non-customers can act as detractors and generate bad word-of-mouth publicity.

Furthermore the NPS disregards important differences in the answer score distribution: no distinction is made between a 0 score and a 6 score, while there is obviously a substantial discrepancy between those two. It also makes no difference whether there are 70% Promoters and 30% Detractors or 40% Promoters and 0% Detractors. Both result in an NPS of +40 which doesn’t seem very logical.


Some caution is therefore required. It is obvious that the NPS on its own may not be sufficient as a management tool, but in the right framework and with some additional motivational questioning it can undoubtedly be a useful metric. Its greatest strength is the simplicity, making it functional and interpretable for every stakeholder. Considering the large number of big companies using this tool on a continuous basis, it certainly has proven its worth.

To be clear, NPS is only a starting point. After the analysis, the real work can begin: improving your organization and taking actions to boost your NPS. This will be a long term project, but the NPS allows you to perfectly assess at which stage your organization is in this growth process.

“Net Promoter” is a registered trademark of Fred Reichheld, Bain & Company and Satmetrix.



  1. Hi, I work for a company that constantly presents NPS as a major “weapon”.
    The this is that the whole group uses it as a percentage. I’ve seen that several websites use it as a percentage as well… In your situation you say it is not a percentage; what is your comment on that?

    Frederico Santos
    • Frederico, thanks for your comment.
      Both options (with or without %) are commonly used. Wikipedia states explicitly “the NPS is not a percentage but some people wrongly put a % sign after it instead of correctly using “+” or “-” to show the result”.
      Although you calculate NPS by subtracting the % of Detractors from the % of Promoters, the result is not itself a percentage, but a (positive or negative) number.

      • What if the NPS is negative, what is the implication?

      • A negative NPS means you have more detractors than promotors. Next step is the detection of the underlying problem(s) and setting up an action plan to improve your organisations NPS. You can try to decrease the nr. of detractors, increase your promotors or a combination of both.

  2. If you are going to use a NPS question in your research, don’t forget to add a reference to Satmetrix. “Net Promoter Score” is a registered trademark! Adding a line as we did at the bottom of this article is sufficient.

    Alexander Dobronte
  3. What are your comments on this PR? where the score is represented in % with 2012 industry benchmark

    • Seems they made the mistake of copying and pasting a news article without editing the text.

      NPS as a percentage makes no mathmatical sense in my opinion. Once you subtract the two percentages the denominator looses its logical value [if you find -4% you should ask youself minus four of what?].

      Good discussion though.

  4. I have a query, why is 5 not considered in the NPS calculation

    • the 5 score is included in the Detractor group, as this includes scores from 0 to 6

  5. why is rating 4 to 1 on a 7 point scale considered dissatisfied

    • Some companies use their own variant of the NPS, but for the original 11-point scale NPS the Detractors group scores 0 to 6 and Promoters 9-10. A 4 score on a 7-point scale is more or less comparable with a 6 on the 11-point scale, both in the Detractor group.

  6. What is so special about such a tool? Has it been effectively proven?

    • As with many loyalty or satisfaction indexes you always have a group of believers and disbelievers, the NPS is no exception on this rule. However it is a very commonly used index in many industries, which remains popular because it is well marketed, intuitive and easy to understand. You can easily compare your companies score with other benchmarks and track evolution in time. Whether the NPS is an accurate predictor of future growth remains an (interesting) discussion.

  7. Can someone tell me how many responses as a minimum I should use as a base for calculation. I want to compare quarter to quarter and draw the line under a certain number so I compare like number with like number. I have no idea yet how many I’ll get. I’m hoping to use 100 but if I only get 50 is that enough to draw from?

  8. I would try to go for 100 respondents, 50 is rather limited as a base to draw any relevant conclusions. As with every sample, the higher it is the the smaller your margin of error will be.

  9. Many thanks. And thanks for such a clear post by the way. I’m eager to give it a try. It will be for a product via a pop up on screen. Can I do it as simply as using the same collection method and same number of respondents each time?
    One more question- I assume it’s ok to ask for optional contact details so I can follow up with detractors and maybe promoters?

  10. Bernie,
    Thanks and good luck with the data collection. I would try to go for the same collection method each time to avoid data bias, but if you can extend your respondents sample that’s no problem as this will make your data more accurate.
    It’s a good idea to ask for contact details, but don’t use a compulsory question as this could impact the response rate.
    If you work with the CheckMarket tool to collect the responses, it is possible to receive automated alerts (via e-mail, an sms text message, Twitter message, …) based on a set of conditions (e.g. only from detractors). On our site you can find more details:

  11. thanks there
    how many questions to ask?

    • You just need to ask the one and only NPS question:
      “How likely are you to recommend company/brand/product X
      to a friend/colleague/relative?”

      • If all you want is to calculate a score, you only need one question, just as Gert points out. But as the creators of the Net Promoter system, we never intended (or recomend) you only ask one question. Rather, you must at minimum also ask the customer why.

        The essence of the Net Promoter system is what we at Bain call high-velocity closed-loop feedback, learning and action. The simplicity of the single-question-based score is supplemented by open-ended comments from the customers. All this is delivered straight to people in the organization who can learn from the feedback and act on it.

        For a more complete description of the Net Promoter system, including the role of the survey questions, there are many resources at

  12. Rob, I totally agree with you on that point. The reasons behind the NPS are equally important (or even more important) than the NPS score itself, and based on this feedback an appropriate action plan has to be developed which leads to an improved NPS.

  13. If you share the scoring model in advance with the customers so that they understand how it works, would this invalidate the results as it could be suggested that you are leading them to the result you wish to get?

  14. Interesting question Lou. In my opinion this could indeed lead to some distortion as respondents might try to avoid the Passive answer option (score 7-8). Your NPS index probably will be biased If you want to compare with other NPS scores in the sector, so I therefor would suggest not to share the scoring model with your respondents.

  15. The company i work for have just introduced nps. what do you consider to be afair target to achieve bonus..

  16. It is difficult to set a specific target, because this can differ a lot between industries and countries (some cultures tend to give more positive scores than other). You can search the internet for benchmark scores in comparable countries or sectors. A better solution would be to define bonuses based on the actual evolution of the NPS score compared to a zero measurement you perform at the start of the evaluation period.

  17. We conduct Transactional and Relational NPS surveys. We usually conduct Transactional surveys when customers are leaving our stores. But I´d like to know how long after customers have made a purchase (or other transaction) at our stores can we survey them if we want to obtain a Relational NPS?

  18. Dear Gert,
    thanks for an interesting post! I’m going to start survey on our sites, though I’m not sure how to sample. We have about 250 000 active users on site, so would it be right to totally randomly get 300 answers each month (I suppose I should show survey to about 3000- 3500 users) and next month do same thing but with another random 300 users? Am I thinking right? Thank you very much for your answer!

    • Ingrida,

      What you suggest seems a logical way to proceed. I would try to go for 400 respondents per wave to reduce the margin of error to less than 5% (
      If you expect a response rate of 10% you should send out your survey to a sample group of 4000. Make sure to invite a new batch of people each month (don’t send invitations to users who already received one).
      After the first month you will have a better view on the response rate and you can adjust the sample size accordingly.
      I wish you lots of succes!

  19. Is it possible to make an equation that works for this? My company uses NPS on an individual level. While individual Salespeople get their NPS “metrics” on a daily basis and have a monthly goal to maintain, getting incentives based on performance. I am trying to make (in Excel), and equation that will say: If you earn X amount of Promoters, you will be at your 54% goal. Given my math – it is flawed as the numbers aren’t really percentages so the math doesn’t work like a percentage. What equation could I use to say: Promoters P, Detractors D, Passive V, Goal NPS 54 . Clearly increasing promoters also increases the total surveys and it is not as simple as promoter + x.
    Is there a known way to do that without plugging in extra surveys until my number hits the 54 result? If you have one, I would greatly appreciate it. My math is flawed somewhere. My assumption being that the equation isn’t really a fraction/percentage so won’t act like one in reverse either. I am finding it easier to motivate those below goal to show them how far they are from reaching a number many of them just don’t understand as well as I do. Thanks so much!

    • Hi Amber,

      Interesting question, I have given this some though but can’t find a straightforward solution for your problem. The NPS can be a result of an endless number of combinations of promotors and detractors, and also your denominator is variable taking into account the passives. In my opinion it is not possible to put this in a simple excel equation.
      But if anyone else out there should have a solution for this, feel free to share your thoughts with us!

    • Hi Amber,

      It is possible to calculate this if you know all of the data, or at least the NPS and the sample size.

      NPS = 100 * ( P – D) / T
      where T = sample size = P + D + V.
      So if we get x new Promoters,
      New_NPS = 100 * ( P + x – D) / (T + x)

      Goal is to have the New NPS = 54.
      after some calculations, ends up as:
      x = (T * Goal / 100 – (P – D)) / (1 – Goal/100)
      = T/100*(Goal – NPS) / (1 – Goal/100)

      Example: 18 Detractors, 72 Promoters, 20 Passive
      NPS = 100 * (72 – 18) / 110 = +49.1
      x = 110/100 *(54 – 49.1) / (1 – 54/100) = 11.7

      So if the next 12 people are Promoters, they’ve exceeded the +54 NPS goal. Of course, each Passive or Detractor makes it harder to hit the Goal.

      Having a sample size twice as big will make it twice as hard to hit your goal starting at the same initial NPS. So it would take 24 consecutive Promoters to hit the goal if the initial sample was 220 people and the NPS was still 49.1

      This is the type of stat that might be better to look at a monthly basis rather than a daily basis.

      • Josh,
        This makes sense, thanks for helping us out here!

        Gert Van Dessel
      • It looks like for every 3 new Detractors you need 10 or more new Promoters and for every 5 new Passives you need 6 new Promoters in order to get a NPS of 54.

        P:V = Goal / (100 – Goal) = 54/46 = 1.17
        P:D = (100 + Goal)/(100 – Goal) = 154/46 = 3.347

  20. I’ve seen some studies measure the likelihood to recommend questions (or NPS) based on a past 90 day interaction with the company. What is the standard pratice for this question? Should we ensure that our customers have had an interaction with us in the past 90 days to measure our NPS?
    Thank you!

  21. Sylvia,
    I would recommend to keep the period between the customer interaction and NPS question (or any other survey related to the interaction) as short as possible. Ideally you would have a continuous monitoring system keeping track of NPS on a monthly (or even day-to-day) basis.
    With the survey API options in our tool ( you can even set up automatically sent surveys after each customer contact.

    Gert Van Dessel
    • Hello Gert,
      Thanks for your reply. For the period between the customer interaction and the NPS question, it makes sense when it relates to a transactional NPS to keep the interaction period as short as possible. But would this still apply for measuring relationship or brand NPS? A relationship NPS would be impacted more by brand-related factors such brand reputation, price, product, quality of the product/service, etc. I’ve seen industry NPS benchmarks (measured by the Temkin Group or Satmetrix) not filter based on based on the interaction period. So just wondering if that is the best practice (not using a filter for the interaction period e.g past 90 days interaction or past 6 months interaction)?
      Thank you,

      • Sylvia,

        I do not know your particular situation, but in any case I would still recommend at least one interaction in the last 6 montht as a minimum prerequisite to have a valid NPS score.

  22. I understand how to calculate. Just puzzled that passives are not included yet they are in a way as if you have 10 survey and 6 give you a 9 or 10 and 2 give you 7 or 8 passive then the remaining 2 give you 0-6 your score is 40. Surely if they were passive then they wouldn’t be included at all. Technically they affect your score so how are the passive and not included. To be passive and have no affect those 2 survey would have to be removed completely ?

  23. Hi Russell,

    Passives (V) are included and do lower your score, just not as quickly as Detractors (D).

    NPS = 100 * (P – D) / (P + D + V)

    If you “threw out” those two Passives, you’d have a score of
    NPS = 100 * (6-2) / (6 + 2 + 0) = 100 * 4 / 8 = 50
    instead of 40.

    If your total survey size was 20 and all twenty came back with 10 more passives, your final score would be

    NPS = 100 * (6-2) / (6 + 2 + 10) = 100 * 4 / 20 = 20

    which is much lower than the 50 you would get if you did not include all of the Passives. Still higher than the negative NPS you would get if those 10 new responses were all Detractors instead of Passives!

  24. Ok so how do people score -100 is then when they only have detractors or could they score -100 with passives as well. Thanks for your help its much appreciated.

    • Hi Russell,
      The only way to score -100 would be if 100% of the people who respond to your survey were Detractors. If everyone on the survey were Passives, the final score would be 0 (which would also be true if for every Promoter there was one Detractor).

      • If you start out with a negative NPS, Passives will move you closer to 0, but they only way to get positive is to have Promoters.

        2 Promoters, 6 Detractors, 2 Passives.
        NPS = 100*(2 – 6) / (2 + 6 + 2) = -4*100/10 = -40
        Get 10 more Passives:
        NPS = 100*(2-6)/(2+6+12) = -4*100/20 = -20
        Then get 10 Promoters:
        NPS = 100*(12-6)/(12+6+12) = 6*100/30 = 20

  25. Hi I work for bskyb eg sky tv , we have now been told our payment now will rely on nps is this normal I am a engineer and I can only control the work I complete so I don’t see how I could be responsible for the entire nps of the customer so eg if the products are poor quality or the menus is hard to use or there is not enough content shown or its to expensive or hard to book a instal or service call for the day or time required or the sales team promises something or withholds information then this is now my problem ,I think it’s very unfair to have my pay rely on this ,I don’t mind kpis or a manager tracking them but wages should not come in to this ,any advise ,thanks

  26. Hi Bob,

    I can understand your employer when he wants to get the whole team involved in improving the NPS and company performance. But obviously you only have a minor impact on this score, so I agree with you that the impact on your salary also should be marginally.

  27. Yes I would be happy to have a small bonus on the nps and hopefully I could help to build a good nps for the company but as engineers we will be forced to ask customer to lie and give us good nps to protect our pay and if the customer likes us hopefully they will but this will then lead to a unrealistic nps of the company ,everyone is very angry with this new pay structure and we now feel like it does not matter if we do a ok job and keep the customer happy or if we do an exceptional job as our work quality does not have much impact now its down to nps.i feel sorry for the impact of the customer in the near future ,I hope it does not come to that but I can see it happening ,I’m only a small part in a big company but I always used to feel like I made a difference ,but you are right I do have a minor impact on this score . Nps should be used but not abused

  28. I have a puzzling scenario that I’m looking for insights on. I’ve been doing NPS for a variety of high-engagement sites, and sampling visitors randomly who visit the site using a notification bar at the top of the home page.

    Across my surveying, I have had larger and smaller sample sizes, based on my clients leaving the survey links up longer and shorter times. I got curious about the impact of sample sizes, and wondering what the sweet spot was for sample size. I’ve plotted the “rolling” NPS score – recalculating the score after every new data point, and then plotting across the x-axis.

    What surprised me was that after levelling off after several hundred data points, the score then gradually erodes systematically over time in many cases. I would have thought that the score should stabilize further, but in fact, it just slides down.

    i have my own theories, but would love to hear yours. Has anyone else looked at rolling NPS over the course of an individual data set?

    Brad P
    • Hi Brad,

      A lot depends on how you ask and where. I’ve noticed more and more NPS questions popping up when visiting a website.

      The first time this happens I’m likely to give the site a high score for asking my opinion, but if it asks again on a subsequent visits or many sites I visit have this, I’m likely to give a lower score just out of annoyance.

      Also, are you asking “why” and following up, as Rob Markey points out in his comment above?

  29. Hi Josh,

    I’m using a “notification bar” on the top of the browser window, typically. In the case that’s puzzling, I left the notification up for only 2.5 hours, yielding over 1500 responses. This is an actively engaged site. The first hour, the results, largely matched previous surveys for the same site. The 2nd hour and a half, the score systematically eroded.

    The survey is set to not allow the same user to respond, so if that’s working properly, the annoyance rating shouldn’t be a factor.

    My theory is that on a site with high engagement. you get a selection bias early in the survey, as the heaviest users statistically outnumber the casual users in regard to the “random” sampling. Then, over time, I now believe that the more casual users are contributing – users who are not as enthusiastic or heavy users, and therefore reflect a less optimistic view of the site.

    Anyone else ever look at the data on a rolling basis over time? Would love other stories here.

    Brad P
  30. Ummmm. OK. Well, upon closer inspection, I’ve had a data error. In fact, my results are OPPOSITE what I reported… , in this case the score actually improved over time. I had my data sorted incorrectly over time.

    Looking for newer theories.

    Brad P
    • I bet the early returns are just statistical noise cause by the small sample. I doubt detractors are distributed as early respondents (unless you had a bunch of referral traffic early in the survey from a site with a higher propensity to refer detractors to you).

      • This seems especially true given that–contra your earlier assertion–the scores increased to match previous scores as the sample increased.

  31. There is an error in your article. Absolute numbers are always positive, which makes the following quote an oxymoron, since it cant be negative.

    “The NPS is not expressed as a percentage but as an absolute number lying between -100 and +100.”

    Christian Lindeberg
    • Christian,

      Thanks for the remark. We use the expression abolute number in this case to make a distinction with percentage, not in the mathematical sense of absolute value of a number. The term absolute number is also used in forums and blogs on Satmetrix website (developers of the NPS).

  32. Hi Gert,

    I was wondering when the best time is to ask the NPS question for a services site. Do you ask on the site directly, or afterwards via email? Why is one preferable over the other?

    Thanks !


    • Hi Sasha,

      I would do this on the site with a pop-up appearing immediately when your customer leaves the site. This will increase the response rate and you will get more accurate comments this way, when everything is still ‘fresh’ in the memory of your visitor.

  33. Hi Gert,

    another question re: percentages. Example scenario: you measure NPS at 2 different dates: +10 NPS at one date and +20 at a second date later on.

    Would you / could you say:
    a- NPS increased by 100%, i.e. it doubled
    b- NPS increased by 5% (a 10 point increase on a 200 point scale)
    c- NPS increased by 9% (120-110 / 110)
    d- NPS increased by 10 NPS “points”
    e- none of the above

    Thanks for your input.

    • Hello Max,

      In my opinion option d is the only valid statement you can make about the evolution of the NPS. In the same way you can say an evolution from -5 to +10 is an increase of NPS by 15 points.

  34. Hi,

    We have just started to use the NPS within my organisation. In addition to 0-10, we offer customers the opportunity to answer “don’t know”. When analysing the results, should customers who answered “don’t know”, be included within the detractors sub-set or excluded from the results altogether?


    Sharon Chahal
    • Sharon,

      I certainly wouldn’t categorize these respondents as detractors. If you would have to put them in a category I would call them passives. As this would have no impact on the calculation of the NPS score the best solution is to exclude them alltogether.

  35. Hi, Very nice article explaning NPS in detail.. I just have two questions:
    1) What is the minimum percentage of samples required to be taken for an organisation of 150000 customers to get a valid response
    2) Does every transaction response NPS transform to the final relationship NPS?

    • I mean will 500-600 samples be sufficient to calculate NPS?

      • Hello Neeth,

        I would go for a minimum of 500 samples, but I would repeat this on a monthly basis to track the evolution over time. On yearly basis you would then have a very solid base of 6000 responses.
        Can you elaborate on your 2nd question, this is not totally clear to me

  36. Thanks for answering questions about NPS! I am wondering about errors with the Likert Scale. For example, a customer gives the highest marks on all questions in the survey but then at the end gives a score of ’0′ for likelihood of recommending to a friend/relative. Could this be a mistake? If so, would I still count it as a ’0′ no matter what?

    Consuelo Gutierrez
    • Consuelo,

      If you think this is not correct based on the other answers I would leave this one out or consider the respondent as a ‘passive’.

  37. Great article,
    If we have been using the standard 10 pt Likert scale without zero for likelihood to recommend, do you think that our NPS results will be problematic? From a trending perspective, I believe it is the same result – movement in the metric will be the same as if the scale included 0.

    • Chris,

      As you have less detractor answer categories, you should expect the NPS score to increase. But in this case the impact this will only be marginal (if any at all). Someone who wants to give 0 score will now give a 1 score, which will make no difference in the end for the calculation. It is however important to continue with the same scale for the trending.

  38. Thanks Chris!

    I have one more question. We also use the NPS for internal ratings of departments by fellow employees, it looks to me that many people give a score of 5 or even 6 when they don’t have anything particularly bad to say or when they don’t have any interaction with a particular dept. I’m going to assume these are passives rather than detractors. Would this be a fair assessment?

    Consuelo Gutierrez
    • Consuelo,

      I would be very careful about giving your own interpretation to the answer scales and changing the intervals. Just stick to the standard calculation and track the NPS scores over time to see what the trend is.

      Gert Van Dessel
  39. Can NPS come in negative ?
    In my case it’s coming .
    N my data is given below ,

    P: Promoter (11.89 %)
    N: Passive (40 %)
    D: Detractor (48.10 %)
    Hello , my data is this now how to calculate NPS.

    To Mr. Gert Van Dessel

  40. Yes, NPS can be negative, it varies between -100 and +100

    Gert Van Dessel
  41. Can the NPS method also be used to calculate agent level CSAT scores? My current company does that, but in some other companies I have seen agent level CSAT being calculated as an average score of all their surveys.

    • Many large companies do use NPS as a major (or the only) indication of customer satisfaction scores for their agents, others can of course calculate the CSAT based on a combination of questions, inclusing NPS or not.

      Gert Van Dessel
  42. Good afternoon:

    I have some questions. Is helpful to explain to the interviewee what is being detractor, passive and promoter? What are the number ranges? and in our case, how can we analyze the drivers of the service identified?

    Thank you very much.


    Gabriela Merino A.
    • Gabriela,

      I would certainly not explain the interviewee which score makes them a detractor, passive or promoter, as this will influence the NPS score and make this not comparable to other NPS scores.

      The ranges are 0-6 for detractor, 7-8 for passive and 9-10 for promoter.

      If you want to know the drivers for the score, you can add an open question following the NPS question where you as for a motivation of the given score.

  43. The NPS is based on flawed mathematics and flawed questioning. I don’t think it gives useful information –

    • Alex,
      I agree in all it’s simplicity the NPS doesn’t show all the nuances behind consumer satisfaction and that is one of the greatest flaws of this index. On the other hand this simplicity is also the strength of the NPS score, the one number gives you a first indication of your companies performance and as it is widely spread between industries you have lots of benchmarks. An evolution over time of the NPS for your company in my opinion still is a very usefull indicator of the direction your company is going. But as it is the case with every number, you have to look behind the mathematics and try to detect what’s behind it to take appropriate actions.

  44. hi gert,

    im a reporting analyst and i just want to know the best formula for nps%togoal. for example:

    target = 45
    survey scans = 100
    promoter = 34
    detractors = 18
    passive = 10

    thanks for your help.

    • Hello Jhoey,

      The NPS in your example is 16 (34-18). If the target is 45, you are 29 NPS ‘points’ from this target. As the NPS range is from -100 to +100 I would say the % to goal would be 29/200 = 14,5%, but I recommend to work with absolute NPS numbers.

  45. Good afternoon
    We have a business that is consultation based, ie not e commerce. When is the best time to catch them ie via email or text. We are after a good response rate. Do you have any data to support that text or email is better.
    Thank Lucy

    • Hello Lucy,

      There is no conclusive time to send your emails. The best tip is to consider your audience in your email marketing.

      Optimal launch times are also subjective to the device the recipient is using. Desktop and smartphone users are most active between 3 and 4pm and are more likely to open emails during business hours. Tablet users are most active from 8 to 9pm and are more likely to open emails outside business hours.

      Overall, it’s safe to say your audience is more likely to open your email after 12pm. The most active hours overall are from 2 to 5pm.

      Text messages (SMS) have an impressive open rate of 98%, with an average response time of 90 seconds.

      Start doing some simple email and text split tests and see which times and which medium your recipients respond to best.

      Sam Van der Veken
  46. Hi

    Nice article.. Is it better to use (no) (may be) (yes) for the question, instead of number scale between 0 .. 10

    what are your thoughts.

    • Sathya, the NPS question is a worldwide standardized question type, with use of the 0-10 number scale. For benchmark purposed I advise to stick with the defined methodology

  47. We are a B2B company that has many users from a company using our service. So when multiple users from a company respond to a survey, we average their rating and then use that to calculate our overall NPS. Like five users from a company might have scored us a 9,9,7,6,5, so the average for that company is 7 and we drop that company from our calculations.

    Do you think this is a right way to calculate NPS, or should we use the entire user responses for our calculation?

    • Jagan,

      I would not drop an entire comany based on an average score. An average can hide a lot of heterogenious scores (e.g. 7+7+7 gives the same average as 1+10+10, but the story behind is totally different). My advice is to use the individual scores. For historical benchmark purposes, you can also still keep the ‘company level’ NPS

  48. Thanks for the great article. We’ve used NPS for a while now. I’ve been asked why passive scores are relatively high (7 and 8). My thoughts are that NPS could be though of as a scale between -10 and 10 (-10 being likely to warn people away) but then concentrated into a smaller 11 point scale for ease of use. This would mean that a score of 0 is actually someone who is really annoyed and will warn others not to join, a score of 6 is half way and is not annoyed but would not recommend (hence being a ‘detractor’) and 7 and 8 would translate into 6, 5, 4 or 3 on the 21 point scale and make more sense as a passive score. Do you think this is a fair description of why the top end categories are much smaller?

    • Olie,
      This is an original argumentation, but the easier explanation according to me is that a promotor is someone who is willing to convinve others to use your product or service. If a person gives you a 7 or 8 quote, this means he/she is (fairly) satisfied, but not necessary enthousiatic enough to promote you to his friends or relatives. If someone is giving a 0 to 6 score however this means he/she is not (really) satisfied, and as a result a possible risk for bad word-of-mouth

      Gert Van Dessel
  49. What to do with rounding NPS in graphs?

    We are calculating an NPS and had the following results:
    Promoters: 83 (which is 58,45 %)
    Passives: 12 (which is 8,45%)
    Detractors: 47 (which is 33,08%)

    We show these results in a graph that runs up to 100% and shows figures without comma’s. This means the graph shows:
    Promotors: 59%
    Passives: 8%
    Detractors: 33%

    The graph would than show a positive NPS of 25, but calculated on the original results it would present a NPS of 26.

    To calculate the NPS are the actual percentages used or do you use the percentages which are presented after rounding on a 100% scale?

  50. How often per year should you send out a NPS survey?
    And once you get the results, should you focus on the passives or the detractors?

    • Anna,

      If your client base is big enough, the best option would be continuous tracking, a lot of our clients do this. You can ask the NPS question after certain types of interaction, e.g. when a customer orders one of your products. This way you can immediately follow up dissatisfied clients. With the notifications function in our survey tool you can trigger automatic follow up actions (e.g. send an e-mail to the responsible account manager). In this way you can take action to convert your distractors into passives.

      However you shouldn’t neglect the other two groups neither. Try turning your passives into promoters and keep your current promoters happy to avoid them dropping back to the passives group.

      Gert Van Dessel
      • Hello,

        We want to continually track our NPS and ask at various parts of the customer journey.

        I am confused by a couple of things : -

        1. Can I generate a NPS for different parts of the business? ie one for client support, one for production etc. Then do one for the company as a whole?
        2. Is it running data? So if I report back on a monthly basis, if someone gave us a 2 in February is that still counted in the results in September? Does that make sense? Or do I start from a clean slate?


      • Gemma,

        For Q1:
        If your respondent base is large enough you can calculate NPS for different parts of your business. It’s best to focus on the evolution of the separate NPS scores per department or service rather than to compare them. E.g. your customer complaints department will probably get lower NPS scores because they have to deal with unsatisfied customers, but it(s the evolution that counts.

        For Q2:
        Again if your respondent base is large enough you can track month by month for the short term follow-up, and on long term you can use year by year NPS scores

        Gert Van Dessel
  51. How do you calculate an acceptable sample for your customer feedback CSat: ( if your call center answered 1239682 calls and per agent 830 calls answered however the overal Rating from customer is 2923 and Csat is 91.71%. Is this a true reflection of our customers satisfaction? what would be the acceptable sample as per industry?

    Itumeleng Maimela
  52. When a caller is answering the NPS question(s) through the phone (IVR), the 11 point scale can be a bit tricky if the caller input a 1-0 slowly. It can be taken as a detractor (1) rather than as a promoter (10). To ask if the caller meant to enter a 1 rather than a 10 could be considered leading the caller. If the options were reduced to a 0-9 score, what impact would that have on the overall NPS scoring?

    • I would try to focus on solving the technical telephone problem. If you use the 0-9 scale, you won’t be able to benchmark your NPS results with other NPS scores.

      Your NPS will probably be higher compared to the ‘standard’ NPS if you use 8-9 for promoters and 0-5 for detractors (people are intuitively more likely to give a 8-9 score than a 9-10 score). The only comparison you can make with this adapted NPS is with your own historical data.

      Gert Van Dessel

Leave Comment

© 2014 CheckMarket. All rights reserved. Terms of use. Privacy policy.