Sorry NPS, I’m not buying (it)..

I know a lot has been said about Net Promoter Score (NPS), and I’m not in this world to judge anyone who’s working with it, or developing it into a Net Promoter System. I do like to share my experiences with it though, hoping to attract other people who’d like to share theirs, so we can all get a better understanding of what drives Customer loyalty and how to manage for it. Unfortunately the Net Promoter Score is not working for me right now. And here’s why:

In my role at Delta Lloyd Groep I have the pleasure to work together with Zanna van der Aa, who is working in my team as Program manager of the Customer Experience Program we’ve launched this year. Zanna recently received her PhD based on her research on the role of the customer contact center in relationship marketing. In short: she pretty much knows her stuff and she’s as curious as I am to really understand what drives Customer Loyalty ;) (and she blogged about this in Dutch on here)

How we measure
As part of the program we are measuring Customer Satisfaction, Net Promoter Score and Customer Effort Score. The first two we are both measuring on the level of interactions (e.g. after a service call, or claiming damages) and the level of our annual Customer Satisfaction survey on a large proportion of our Customer base (including those not in interaction/transaction with us over the past year). Customer Effort Score we are only measuring on the level of interactions. Apart from these standard questions, we are asking more questions in different forms, including an open answer box to obtain qualitative feedback as well. Response rates are quite high on the transactional surveys, and very satisfactory on the annual one.

What we see
Customer Satisfaction ratings are quite stable and have been increasing steadily over the past years. The scores are also very similar throughout both methods. We seem to have a good understanding on what needle we need to move to get improvement on Customer Satisfaction. How different is this with Net Promoter Score. The score itself is all over the place. It seems to change from quarter to quarter going up and down without any reason (and we have been looking for them).

I’m ‘bothered’
A recent event really makes me doubt the Net Promoter Score question/methodology: Our own measurement showed a score, whilst a survey held by the same research firm on exactly the same sample, as part of an industry benchmark as little as two months later, produced a difference of 20 points in the score. And Customer Satisfaction scores in both surveys showed the exact same result. On top of this there are even bigger differences between the score in our own measurements and other so-called ‘industry benchmarks’. Since for the latter we don’t know the exact way the questions are asked and in what order, we could not really be bothered. But with current ‘evidence’ that’s exactly what we are..

Oh.. and the judge is still out on Customer Effort Score (CES), but so far we don’t see the higher relationship with Customer loyalty, as promised..

So, what do you think? Back to Customer Satisfaction as the primary metric? 

[Disclaimer: This blog in general and this post in specific reflects my personal opinion only]

36 thoughts on “Sorry NPS, I’m not buying (it)..

  1. Talk with a biostatistician. In addition, in some document I read, the r-squared of the ultimate question is 80%. Coincident? Regardless, of the swing, such can be smoothed out with moving average, examining the standard deviation and etc… Overall, NPS, in my opinion provides a great beacon to help many department focus on what is important to the customer by listening to them. Customer is #1. NPS helps us refocus on what is most vital to a business. My preliminary work into NPS, I was able to identify all the problem child product we developed in the past few years. Wow, that’s a powerful tool. In addition, I was able to unearth products we didn’t even know was awesome. I reviewed over 300 Skus across almost 4000 reviews in a single year…if I had the full social media dataset, I would be a my stats guy would be so happy.


  2. A late thought on the good blog and the interesting reactions. Like many other researchers I was confronted with the wish for NPS by top management just after the credit crunch in my work for ING. Although a lot of my background research showed NPS was not the best Metric for a good future loyalty indication, I implemented a large scale NPS research program at ING. “if you can beat them, join them”. However I always kept satisfaction in the surveys.

    I decided to focus on positive change and not to much on the metric (the metric is a means to an end). However sometimes also for me fluctuations in scores are hard to explain. Nowadays in other companies (as consultant or interim) i still keep focus on effect on change whatever metric. Though I prefer simple satisfaction questions. (Just not mean scores ;-)

    Jorrit Lang


    • I would like to invite those people who are frustrated with NPS to try InfoQuest. We include the recommend question in all our surveys (and have done since 1989, so you have our world-class benchmarking score). But, more importantly, we can pose up to 59 other questions and statements as well before our average response rate of 70%+ is affected. (We only work in B2B, but can cope with all business languages). Please take a look at our library of questions – item 2 on . There is a direct telephone number for me on the home page.


  3. Hello Wim,

    Thanks for sharing your experiences. I kind of steppes out of the NPS debate. However my experience at ING is similair to yours. And currently with my new clients roughly the same thing. Still of NPS is a KPI that can move people in your company it is for a good cause ;-)


  4. hi: i just dont get it — we use metrics to drive decisions — nps or satisfaction — and these metrics are based on assumptions… that are often out of date… metrics are great in static environments.

    I value indicators instead. Think of them like levers. You need a bundle of them, certain mixes of them. Especially early warning indicators (lead e.g. am I asking for more, am I sending in ideas, have you actioned any of my feedback, I am using what your offering in terms of your collaboration content hub) rather than after the fact (satisfaction or nps)

    I know for me personally, as soon as the company demonstrates, sharing, community, care, kindness values…. real genuine.. they will be in my “advocate” group. And that is irrespective of the product, level of service provided.

    for example
    a) is big in this area. There is alot I’d like to see different with this sCRM, but I advocate them currently, because of the support they give non-for-profits.
    b) contrast to Apple. I logged an idea with Apple, that they allow us to gift “our unused apps” for iphones/ipads to friends or a big bucket in the sky. So embrace the concept of re-use, re-distribution. Unfortunately Apple have failed to respond to my suggestions :-(. Now I love their products, but I would not advocate them.

    No metric captures this cause/effect. This cause/effect will increase if “rise of collaborative consumption” continues to gain momentum. That is re-design/ return to different ways of consumption (gifting, sharing, access v ownership, etc..) if your interested to know more, here is link:

    Customers, consumers, me — can work out a companies survey mechanism — and use for a different purpose than intended survey. For example, if I get a survey after a good interaction via email, I’ll rate it poorly. As nothing is worse in being sent something, cause the agent is expecting a good rating. yuck!

    Ultimately, for me, it depends what action we want from the customer.
    – do you want their input, to learn from them about a specific agent’s or specific interaction or value of xx?
    – do you want their input, to learn about the perception of your organisation?
    – do you want them to advocate?

    If i use myself as an testcase and my recent replies to my telco or apple’s survey
    – For me to recommend, meaning rave about, rather than respond if asked, it takes alot. I am satisfied with apple and my telco, but I would not recommend either. I am satisfied as I have no open issues and all issues that have arisen have been handled professional. But recommend. No. I’ve had more than 1 issues, and some are same issue repeated.
    – For me to be satisfied, is a fairly low level. As Paul Greenberg says in his Era of Engagement keynote at sugarcon, “just have to give me what I asked for”. Which hopefully the organisation would always consider reasonable.

    So in my non-mathematical research, what i have discovered is. People who have had a “great experience”, like a “story to tell”, will advocate. so to me its not a score but rather like a story count — 3 great stories.1 great story. This means the majority of customers will be 0 stories. That is not bad. Just means no opportunity yet.


    • Wim,

      I cannot agree with your statement ‘No metric captures this cause/effect….’. The NPS score itself should always be followed by an open question according to Reicheld “what is the main reason for your give score’. We as consultants always advice and execute some additional smart questions in our surveys.).

      With the right questions you get to know why you do recomment sForce en do not recommend Apple.

      Frank van Gool


  5. Wim, all,

    Great debate! Most interesting i find the focus the words NPScore of NPSystem. Where my experience is that NPS can immensely help in focussing on an NPSpirit! Sounds vague but it means in concrete: It’s all about reaching a way of working where the customer is central in all actions and decisions been taken.

    I’m fully confident that you’ll manage to evaporize some of the fluctuations but even we se lots of ups and downs. There are ways to flatten them out but the fluctuations will always be bigger than with customer satisfaction. Taking the WOM into account could help a bit!

    The point is, unless you already get 9’s and 10’s in C-sat, customer satisfaction doesn’t tell you a lot about your customer’s behaviour (een 7,5 gaat bijna net zo makkelijk weg als een 3!), the theory by Sasser & Jones (9+ economy).

    I firmly believe that the NPS as a single metiric is the best available yet! Implicating that organisations shouldn’t only focus about how to flatten the fluctiations, but merely on unerstanding that the NPS shows you the real emotional experience by your customers. With all the outer influences. The mission of a CE-captain therfore mostly is to inspire it’s organisation in understanding that it’s not about getting a target on 1% but all about setting an upward trend (measured in NPS).

    With all the other focus om how to implement and improve with a continuous 2 or 3 level learning loop and gaining and improving customer insights and customer service using emotional customer journeys and focus on your fan’s almost as your main marketing focus (stop advertising!).

    I’m sure someone feels to react on this, please do so as this is more about a belief than about a measure!



  6. Wim,
    Yes I’ve experiences similar gaps in the math. Beyond that, in my experience, the actionable data comes from the follow up question of “why”. So, whether you ask would you, will you or did you recommend, the organization can’t do much with that answer until it understands why or why not. Same applies to csat.

    I did hear an interesting presentation from Safelite Autoglass here in the USA thought that reported on how they are using NPS as an internal quality measure. The challenge with existing quality measures that aren’t reviewed or periodically challenged is you get to a point where your reporting quality scores of +90%. And the identification of opportunities for continuous mprove,ent become more and more difficult. So, whether it’s NPS or some other measure, simply changing the dimensions against which you measure the business periodically can expose hidden opportunities for improvement. If NPS works in that capacity, then have at it.


  7. Relying on just transactional surveys is like an ostrich putting its head in the sand. It really doesn’t matter if you are measuring satisfaction or loyalty (BTW, There is actually a high correlation between satisfaction and other measures of loyal intentions), top box raters still will leave for better alternatives when they exist – remember the walkman, slide rule, 8-tracks, Beta, etc..

    Another good example is AT&T. I worked with them in the early 90’s. AT&T had high satisfaction and loyalty ratings when there were very few alternatives. When the Bellco’s started offering long distance and, even to a greater extent, VoIP started to become real, loyal customers were bailing left and right. Our research showed the competitive treat and what management needed to do about it. Unfortunately, they didn’t listen to the data and follow our recommendations on priorities. They focused on improving service and spent big $ on reinforcing their brand through advertising. It was wasted money because they were focused on the wrong issues. 20 years later, most organizations still don’t get it.

    The main problem is, by and large, that most organizations are way too tactical in their thinking. Annual strategic research is used to take an objective picture of SWOT to get organizations to look forward, consider their performance in a competitive context, and take actions that give them a clear competitive edge. A strategic plan is built on objective, fact rather than subjective, opinion and assumptions. The market-driven priorities are then cascaded throughout the organization; employees are held accountable in performance reviews; and progress is tracked over time. Priorities are set with predictive financial models using the current and prospective customer data to focus on those areas having the greatest return-on-investment.

    Transactional surveys can support this effort, but by themselves can be extremely dangerous. They give management of false sense of security. By the time they wake up, they are usually at a competitive disadvantage and losing market share. For an example of the impact, look at the comparison of a market-leading organization that has the highest NPS score among competitors yet is losing market share and being hammered in the stock market (second graph – It doesn’t get much better then this. I love helping organizations compete against NPS and tactical competitors :)

    All the best,



  8. Wow, this is a great debate about NPS and why you should use it or not. I have a few additional comments.
    First, I do believe that NPS is good metric for customers to measure their intention to share their enthusiasm about a brand, product or service. In that sense it is stronger than satisfaction because the latter is a more passive metric (and therefor more stable?).
    Second, and that is said earlier in the dicsussion, find the proper way to ask NPS that fits your organisation. So adjusting the question is no crime but common sense, as presented in NPS-conferences all over the world.
    Third, the succes of NPS is too strong to ignore, but it becomes a succes if not treated as a metric or score, but as a system (implemented in the business) or even better as a spirit (implemented among your employees). Carglass is a very good example of a company that strongly believe in NPS which they have translated trough the whole business from top to bottom (and up again). Creating a closed loop feedback is vital to create change and more customer awareness in your organisation. Read Reichheld latest book: the ultimate question 2.0.
    And last, do not focus only on your detractors, but focus on your promoters as well. Those are your loyal clients and by giving them attention and listening to them, it will create company pride among your employees, a sense of urgency for delivering better service and it will keep your loyal customers connected to your company and they will recommend you. After all, consumers are more likely to put trust in a neighbour or friend than in advertising of big companies. So engaging your promoters is the real secret of NPS but in order to do so, you have to measure it.


    • You raise an excellent point, Els, about focusing on your promoters. Some of these will be evangelists for you and are capable of generating a healthy number of referrals.


    • Hi Els,

      Thx for the comment and addition to this dialog. I very much agree with you, but there’s nothing in what you say I cannot do (or am not doing for that matter) with the traditional Customer Satisfaction question/metric..

      One thing though: one of the main reasons why Reichheld c.s. have named people, that express criticism on NPS, ‘Net ProMoaners’ is that they supposedly have a vested interest in their own methodology, that they have an ‘industry’ or position to protect..

      With all the NPS conferences, the book, the ‘System’ et all.. the same could be said about people promoting the Net Promoter Score/System, no?

      I have no vested interest what soever. I’m trying to work out strategies to improve my employers Customer’s experience, and NPS measurement is not helping me to get that done..

      What do you think?



  9. Although I think the nps particularly strong driver for changes in the organization due to its popularity and some irrational behavior.

    For me in in an outsourcing context client focus on NPS often drives a stronger motivation for identifiening and solving procesproblems and siloproblems in partnership than a CSAT score. Main

    driver is management focus and scale. If on average you score a 7 on CSAT there is no panic about performance. Equivalent score on NPS normally results in a negative outcome. Negative scores often

    look more dramatic.

    The initial performance improvement based on NPS is often easy. From negative values to neutral values can be largely done by optimizing the “contact” processes and the way the customers are

    adressed. The second phase, often more complicated to get to root causes and to see relational effect of your efforts, deals mainly with an organisation as a whole and the integration of multi-silo


    NPS usesfull; yes but mainly to focus the nessecary organisational change to customer centricity.

    I think there are two main reasons for irrational behavior of this KPI: mathematics and customer behavior.


    Assuming you did your statistical homework e.g. sample size, selection, distribution, target audience and you even in some way take in account the “non” responders.

    The mathematical nature of the NPS scale gives only voice to the extremist: the detractors and promoters. An average customer who gives an average score of 7 or 8 results in a score not attributing

    to your NPS score. In the NPS metric the “neutrals” act as if they are non-responders to your survey. In contrast with a CSAT score where a 7 or 8 does count. In cases where the distribution of

    NPS score has a high number of neutrals only a small sample of your customers contribute to the NPS metric. Say 70% of your responding customers score neutal, it does mean that only 30% of your

    responding customers contribute to the value of this KPI. A smal shift in evaluation of one point kan make you an NPS extremist in the case you change from a 7 to a 6 score of from an 8 to a 9

    score. Within CSAT the avege is hardly influenced, in NPS you become suddenly an attibutor to the KPI value. The larger your neutral-base, the more severe this effect is.

    A small -simple- example of this principle based on 10 responding customers:

    scores (BASIS)
    Average score 7,2
    NPS 0%

    scores (8=>9)
    Average score 7,3
    NPS 10%

    scores (7=>6)
    Average score 7,1
    NPS -10%

    With 70% neutrals in your base score, your average score looks less sensitive to change in extremist scores; only 0.2. Your NPS score however has a range of 20%pt with a flip of sign, looks more

    dramatic to most people.

    The higher the percent of “neutrals” are the a stronger the voice of your NPS extremists is.


    The NPS question ask for intent of behavior. Not for what you have done. Personally, i think there is a large gap between asking what you “would” do in contrast with what you “did” or “will do”.
    Any customer interaction is interfering largly with your “would” answer -overestimating once behavior- while factors for the “did” answer also lie in personality and the personall environment -not accounted for in most surveys.
    At the moment I only have a very small sample of a “did you reccomend” dataset, but first glance gives an indication for a more stable respons.


    • Thx Michiel,

      I think you’re right. And maybe it’s because of the way the math works that NPS is such a good catalyst for change. That’s great of course, unless you continuously need to explain changes/differences like the ones I mentioned in my post…

      Great to hear you are looking into behavioral questions, not intentional ones.

      Thx for your contribution! Much appreciated.



  10. Over the past couple of years NPS has gained prominence in several companies as one of the most important measurements of customer satisfaction. In industries such as Telecommunications it has become the standard against which companies benchmark themselves and have incorporated it in their internal performance and reward systems.

    NPS is a good indicator of customer satisfaction about a Transaction, however it does not say much about the Experience of the customer or the Relationship with the customer. Since NPS measures the response to a question, not the actual behaviour of customers it is not an indicator of customer loyalty.

    A good application of NPS is for instance when a customer scores low after an interaction with the call centre, the organisation follows up with the customer to understand WHY she scored that way. This enables the team lead and the call centre agent to learn what caused the low score and to do something about it through training, process adjustments, communication, etc. This improves the experience in future customer interactions and at the same time “wins back” the dissatisfied customer, or detractor.

    A painful example is where an organisation uses NPS to measure customer satisfaction in the different touch points. The customer gives a positive score after an online sales transaction, but calls the call centre because of a problem with the delivery of the service. The call centre gets a negative score on the sales transaction that actually failed in another channel.

    In the first case the NPS is score is primarily used to identify issues and positive behaviours, and secondly as a trend line to measure performance of teams in a specific process/transaction. In the second case the NPS scores are rolled up to department level and give an inconsistent performance indicator, without any link to the underlying interactions. What’s worse, by tying the NPS score to internal rewards, it encourages silo behaviour where the NPS score might improve, but the customer experience does not.

    The more complex the product/service is from a customer perspective (telecoms, banking, insurance), the more rigor is required to properly deploy and use NPS. Companies need to decide the purpose, objectives and application of NPS, and how it relates to the real customers’ experience. Here are some considerations:

    – Is there a thorough understanding and definition of: Customer satisfaction, customer interaction, customer experience, customer relationship over time, customer loyalty?
    – Do you measure service quality, or experience, or satisfaction (and no, you cannot capture all 3 in a single score)?
    – As NPS is mainly transactional, how will the experience/satisfaction be measured for 70-80% of customers who do not interact with the organisation?
    – How do you operationalize the individual customers’ scores as well as the overall score?
    – Since the customer is the only one who experiences the complete company, what are the relationships between the different interactions, touch points and processes where NPS is measured?
    – How is the NPS score validated/complimented with other measurements to understand customer satisfaction/experience?
    – As touch points/products/services change over time, what is the short and long term use/value of the scores?
    – How are customer facing projects and initiatives incorporated in the measurement of satisfaction/experience?


  11. In my opinion and experience, NPS is quite an useful indicator but it’s being overvalued. Nowadays, many companies are using it as the key metric in customer interaction but they fail to understand the complex interactions between recommendation and other variables such as satisfaction, brand image and switching costs,a mong others. Moreover, NPS is perceived by many as the cornerstne of consumer loyalty, whereas it is, “per se”, an indirect outcome. I also agree about its high volatility.


    • Hi Miguel,

      I very much agree with the label “indirect outcome”. Fortunately more and more companies are treating it as such, and measuring (and managing) other metrics they can influence directly and that have an effect on the score.. We do that too and find these metrics have a better relationship (=higher predictability to) with satisfaction scores than with NPS.

      Thx for your comment!



  12. Great article Wim. I think using just one metric is dangerous. The danger with NPS is multifaceted. Any score needs to be understood in context, which is why you need to run it alongside or as part of the Customer Satisfaction Survey and other metrics. I think it has its place, however it isn’t the answer to everything and certainly isn’t the same thing as Customer Experience (as some people would have you believe), but simply part of it. Another tool in the bag. I wouldn’t kick it out of bed quite yet, but the psychology behind it certainly does need to be understood better.


  13. Hello Wim,

    The difference is scores may be to do with the Cassandra Phenomenon. The following is from Sat%20Surveys%20-%20Best%20Practice%20Guide.pdf

    The Cassandra Phenomenon sits atop the list of biases and influences that undermine the candor and accuracy of a customer satisfaction survey. It is defined as –

    In a survey environment where the respondent believes their identity is or may be known, a strong positive bias filters into responses.

    In a survey environment where the respondent believes their identity is or may be known, a strong positive bias filters into responses.

    The problem is simple. When customers believe that their identity is or may be known, roughly 70% of the population will hesitate to openly voice a strongly negative opinion for fear of the potential consequences. That hesitancy is driven by concern over possible reprisals, the “hassle factor” of potentially being asked to explain or defend such comments, even misgivings about getting a business partner into trouble (however deserving) and thus detrimentally altering whatever relationship is currently in place.

    It’s ironic when you think about it. In most business settings, success is based on the development of personal relationships. Over time, people become acquainted, familiar with their business to business contact people, perhaps even develop something of a friendship.

    When they do, they tend to become more careful of hurting someone’s feelings or expressing views that may get someone into trouble. In other words, personal relationships tend to suppress the expression of dissatisfaction or unmet need.

    The Cassandra Phenomenon does not apply to everyone. Through long observed behavioral patterns we see that approximately thirty percent of the general population suffers no misgivings about clearly and openly expressing their views – both positive and negative. Those folks will look you in the eye and tell you precisely what is on their mind, regardless of how they think you might react to it.

    The problem is that the remaining seventy percent of the population harbours subconscious concerns about the ramifications or consequences of openly speaking their mind. For those people, the only safe approach, assuming they can’t find a way to duck the survey in the first place, is to “soft-sell” or “back-peddle” on complaints rather than openly voice them or have them committed to paper, where the possible consequences cannot be measured or predicted.
    And, of course, for both the interviewer and the company using the resulting data, there is no possible way to know – before, during or after the survey – which portion of the population any single customer belongs in.

    The only possible way to mitigate these influences, and to promote the candid expression of accurate opinions, is to provide survey participants with a vehicle that will provide them a comfort level to openly express their views without fear of consequences.

    How many of the available survey methodologies do you think provide for that need?

    When customers believe their identity is known, 70% will not openly voice negative comments for fear of producing unwanted consequences.

    And there is an 8-page review of NPS (I think you may already have seen this Wim) on the Downloads page


  14. “Reason why is 1000 times more important than the NPS figure itself” (uit de mond van Fred R. himself, 10 mei Breda)

    Daarnaast is bij transactionele NPS het aantal respondenten vaak kleiner waardoor grotere fluctuaties eerder optreden.

    Maar het gaat om het waarom. Als het onderzoek het waarom niet zelf oplevert, zul je contact moeten opnemen. In zowel goede (promotors) als slechte gevallen (detractors). Bij Apple stores doen ze dit zelfs elke dag (daily download)

    Frank van Gool

    PS Er is een grote correlatie tussen hoge NPS en en hoge gem KTV (>8)


    • Hi Frank,

      My main issue with NPS is that I cannot explain the high fluctuations in the score, not that we cannot explain why people rate us high or low.

      My examples are not related to transactional NPS measurement. And I can ask the “reason why” of high/low Customer Satisfaction too.. I don’t need NPS for that. We do, btw, see that the factors Customers claim to be reason for (not) recommending us, show very similar correlations with both NPS scores and Customer Satisfaction scores. Another reason to just stick with Customer Satisfaction..

      Thx for your comment & stopping by.



  15. I like NPS for a few reasons but agree that it is not a silver bullet, nor should it be used in isolation.

    I agree with Glenn, NPS is great first step and is fantastic to unify effort and customer focus across departments in a business. I also like that I can get some (not entirely reliable I agree) benchmarking across a variety of industries as well.

    I think the exciting thing for us will be to correlate NPS to segments and interactions/events as we capture more data across the customer experience.

    An example of this is that as an online retailer we survey post order arrival but will soon be surveying after any returns as well. Pre and Post-Return comparisons and the qualitative feedback we’ll get with those surveys will really highlight where we need to improve things for the customer.


  16. Wim
    You make very valid points based only on experience but also looking at it from broader context. I am quite surprised to learn that Reichheld labels is critics as Net ProMoaner. Well that sidesteps the argument based on material content and makes it personal.
    Since you already made some of the key points on customer satisfaction metric and how NPS is all over the place, I will try to make other points.

    1. By the very definition of the metric, when you measure it for the first time for any business you are highly likely to find negative value. Take a case where respondents all answer randomly and avoid rating 0 or 10. The NPS score for this business is going to be -55 (whether it is uniformly or normally distributed may change this, but still a low negative number). On the other hand a simple average of the rating (like customer sat) will show it is 5. A negative number creates a sense of really bad situation and helps to sell add-on services to improve this score.

    2. What is materially different between a 10 point scale used by ACSI and this 11 point scale? Why 0 to 6 are detractors when 5 is labeled on NPS survey question as neither likely nor unlikely.

    3. If we treat 7 and 8 as Passives, what about those who did not even bother to respond to the survey? By its own logic this scale should treat all non-respondents as Passives. It cannot switch scales for its own convenience.

    4. When you ask respondents a rating question, regardless of whether it it about satisfaction or recommendation, the answers are not going to be that different. It is an interval scale with continuous data. But in order to brand and monetize it they invented this complex Net math that converts customer level data into a useless company level net metric.

    5. Segmentation? The method and metric absolutely fail to ask any question about segmentation and do not care how this metric differs across segment. What is the use of aggregate metric that ignores segmentation?

    In essence we have an arbitrary irrelevant metric that has captured the minds of some.


  17. @Chris Bailey: Excellent analogy about the doctor and blood pressure. I know that there are lots of critics of NPS. Where I think it’s valuable is for those organizations who have never been customer-centric and are trying to evove, NPS is a great first step; attractive in its simplicity. As I said in my first comment, most companies that I know of who are using it, ask other questions to either widen the discussion or verify the data.


    • Hi Glenn,

      I’m not a critic, nor a Net ProMoaner (did you know that’s how Reichheld names NPS critics in his latest book? bit of a downer if you ask me, but that’s not my point here).

      I’m all in favor of simple metrics and that’s why we also measure NPS. I do like my surveys to produce actionable insights. At this point in time NPS just isn’t doing that for me. Nothing more, nothing less. I’ll continue to use it, hoping for better times, and maybe against better judgment ;)

      Thx for joining in the discussion.



      • “I do like my surveys to produce actionable insights.” For me this is the key point. If actionable insights is the goal, then uncovering why a customer would promote or not seems to me the most valuable part of the NPS system.


  18. I agree with txglennross that there might be other factors that account in your score. E.g. in insurance variances might be explained by customer confidence, politics (e.g in health insurance and life insurance) and a bad or good press.


    • That’s indeed a very good point Fred (and Glenn of course)

      Unfortunately, if what you say is the case (and I think it’s very plausible) NPS is a nice metric to follow, just not one to steer your business on since it will be highly unmotivating to improve and not see the needle move, I think.



      • Again, you’re just using one metric to steer your business. It’s like only using monthly profit to determine whether your business is healthy or not. Yes, your profit may be up for April, but your employees are starting to leave in higher numbers. And those profits may be boosted by a recent acquisition.

        A doctor wouldn’t just look at blood pressure to determine overall health…I’m still trying to figure out why businesses think it’s possible.

        What am I missing?


        • Hi Chris,

          I did not get around answering your first comment, but I’m with you! I fully agree with you that one metric is not enough. I use a balanced scorecard of different metrics, and we are continuously trying out new metrics. On top of that I’m also using other methodologies, like Customer Journey mapping, Customer communities/panels, behavioral analytics etc etc..

          Combining insights from all of these usually gives you a very strong sense of what needs to be improved and what first.




      • I do not believe in the rationale of defining one final big question.

        But bear in mind that as a metric it still has relevance.

        E.g. in a world in which social recommendations are become a fact of life. And no do not underestimate that NPS enables professionals and organizations to unlock thinking with more creative processes.

        If this implies a need for motivating, I do not know. In the end it is about doing your business more effectively.


        • Frankly, I don’t see the metric unlock any creative processes. I believe it’s treated just like any other metric and more often than not, in combination with other metrics.

          Unless you believe that, what I see most, asking to ‘like us on Facebook’ is a creative way to run your business more effectively ;)

          Thx for commenting Fred. Much appreciated :)



  19. We don’t use NPS in my organization, but I’ve been at conferences where speakers used it in theirs. Nearly all of them asked other questions. I agree with inspectiv about over relying on one metric.

    On the other hand, Wim, is it possible there are other factors accounting for the variance in your score? (I’m guessing not, given your experience, but I always prefer to ask, rather than assume.)


  20. Wim, I’ve never been convinced that NPS is the right metric in which to understand customer experience. Might be because it tries to distill a lot of humanistic feelings and beliefs down into a singular question. Perhaps that’s why it’s all over the place…because it tries to accomplish far more than it can and should.

    But here’s the question I have to your question at the end: why does there need to be just one metric driving everything? Are we forsaking intuition and in the process finding more ways to lie to ourselves about the customer experience when the singular point of data doesn’t make sense?

    I don’t ask these questions of my own because I have the right answer. I ask because I’m sensing the pendulum moving far to the side of analytics (i.e., the perfect analytical model will solve all our problems) and am beginning to wonder if we need some sort of rebalancing.


  21. Pingback: Customer Service Agent Collaboration Helps Move The Needle On FCR and Customer Satisfaction | Forrester Blogs « Serve4Impact

Comments are closed.