Net Promoter Score: Four Problems, Two Remedies


Share on LinkedIn

Net Promoter Score or NPS has been around for about 18 years. It’s a simple way to get a pulse on how your customers are responding to your organization, or their loyalty as called by the authors of NPS. The NPS is a formula based on “How likely is it that you will recommend us to a friend?” based on a score from 10 to 0: Customers responding 9 or 10 are called “Promoters”, those responding 0 through 6 are called “Detractors”, with the remaining customers at 7 or 8 called “Neutrals”. By subtracting the percentage of Detractors from the percentage of Promoters, you have the “Net” Promoters. NPS ranges typically from the 60s to low 80s (USAA, Amazon, Bought by Many, Safelite Auto Glass, and others) to minus 15 (some cable and Internet providers)

While there are many enthusiasts for NPS, the trouble is that there are 4 main problems with it: (1) NPS doesn’t answer the critical “Why?” question, so that you can take appropriate actions; (2) You need to be mindful of regional & segment differences; (3) It tosses out the middle of the pack; and (4) It’s not measured often or broadly enough. 

I’ll review these four problems and then tee up two remedies: (A) Predictive Experiences and (B) Effort and other metrics.

Problem #1: NPS Doesn’t Answer “Why?”

The first problem is that most companies stop at the point of collecting the Net Promoter Score. Let’s say they have a 36 NPS across their company with five regions or five operating units. They might say, “Well, last year was a 32 so wow, we’re doing really well. We’ve come up from 32 to 36 with more Promoters. That’s great.”

What they miss and what really needs to be done is to dig into the reasonswhy the score went from 32 to 36 (or if the NPS moves in the opposite direction, why it may have gone from 40 down to 30, because that is often very damaging to the brand).

The first thing is to figure out is why are we getting this NPS, and how have the changes occurred? Ask a series of follow-up questions like:

  • “Well, why is it that you’re giving us a better score?” or “Why did you give us this score?
  • “Why did you give a 5? (high-end detractor score)
  • “Why did you give us an 8?”

Find out the reasons why and then dig even more deeply. Go through the classic 5 Why’s of Kaizen and other types of root cause analysis and then you finally get to some key observations. That’s the first big thing to do is get down to the why’s behind the score and understand the differences from those individuals’ points of view.

Problem #2: You Need to Be Mindful of Regional & Segment Differences

The second problem is that NPS usually misses differences by region, product division, or customer segment, and the ranges can be very wide. You could look at, for example, your best customers, the ones that have bought the most or added the most from an influential or social perspective and find out what their NPS is versus customers that are new, or customers that have been difficult with you and maybe they’re not as profitable for you. Figure out what the difference is in their NPS because if you have a higher NPS with your less important customers or your more difficult customers, that’s a problem because they’re not going to be influencing other people. On the other hand, if you have a higher score with your best customers that could be a very good indicator of success.

You should also examine NPS regionally or by divisions. We had a client with wide variation in the NPS by division — same exact services, but the regional presidents have some different ways they can set up their organizations and their products and their pricing and so forth. But generally, it’s the same. Now, everyone is scratching their heads and we’re helping them to dig deeper by saying, “Well, why is it that the numbers range from 44 in one region, down to 18 in another region?”

Problem #3: It Tosses Out the Middle of the Pack

The third problem is the basic mathematics of NPS. As I noted (and many of you already well know), you take everyone that scores 10 or 9, the top two boxes, and compare to those scoring 0 through 6, essentially ignoring all of the customers who provided scores of 7 or 8.

But isn’t this really dangerous? The sixes and fives are high “Detractors” but basically are in the middle of the pack, so you need to dig even more deeply to find out what’s going on since a little higher and they become Neutrals, improving the NPS, and going lower they could become more serious negative influencers.

The seven and eights could go either way and if they are on that fence, you need to find out who they are and address them immediately but, as noted, their scores are tossed out; NPS pundits and analysts forget about these customers because the math skips them. However, these customers are very, very important: they are influencers in their own right, and they could go up to a 9 or 10 which is good or they could drop down to a lower score, which is bad. You’ve got to figure out what is it that might motivate them, might actually give them a greater incentive to go up, meaning to give you a better score on a frequent basis, on an ongoing basis.

Problem #4: It’s Not Measured Often or Broadly Enough

The last problem is that NPS is merely a sample, and like all surveys suffers from a low response rate, often in the teens or low 20s. This means that 75-85% of your customers never even provide scores to you, and you have to extrapolate from the limited sample size to the entire population. 

While some NPS surveys follow customer interactions, most of them are administered annually or perhaps semi-annually, and yet that’s simply too long to wait given the high velocity in most organizations in terms of product delivery, competitive dynamics, and pricing changes. NPS needs to actually be an evergreen, ongoing analysis — preferably by segment, by region, by operating group — rather than as a mere pulse check: You need to figure out how the patient (your customer) is really doing.

Fortunately, there are two clear remedies to these four problems with NPS.

NPS Remedy 1: Predictive Experiences

As I shared in an earlier column in CustomerThink1, it’s far more insightful and actionable to extract customers’ experiences across “the journey” from discovery through purchase, use, and returns, and thereby predict what they would say if you did ask them about their satisfaction or if they would recommend your services to a friend. For example, you can follow these four steps:

  1. Collect and assess wait times, performance statistics, or completion rates;
  2. Dip into a recommendations engine to select an appropriate action to fill gaps or restore the journey;
  3. Track subsequent performance or actions (including any social posts, survey responses, new purchase rates);
  4. Assess the success of those actions and keep them for future use in the same situations, or select and apply different actions (repeating step 3).

You can also ask customers if they ever did recommend your services to friends (including “Who?” “When?” and, of course “Why?”) and, on the other end, ask new customers or those whose purchase rate has expired if someone recommended that they buy or buy more (including “Who was it?” “When?” and, again, “Why?). 

While this might seem like a lot of data to ingest and process, have no fear! This new age of Big Data that can mask up many data streams, create powerful models, and test hypotheses in near real time is coming to the rescue. Predictive Experience models and solutions are beginning to appear, some completely automated akin to pricing optimization solutions.

NPS Remedy 2: Effort and other Metrics

NPS is supposed to be correlated to and have a causal effect with increase spend/increased sales, which in turn is very tightly coupled with loyalty. However, in the last five years or so, there’s been a new wave of thinking and it’s all around customer effort and making it easier for customers to do business with you.

In response to the NPS question, “How likely is it that you would recommend us to a friend?” the customer has to say, “Well, let me see, I expected a quick response or I had hoped that the product worked the first time I tried to use it, and yet the company was slow or I had to spend extra time getting a tech back to my house, and overall they did OK so I’ll give them a 7.” A Neutral score, tossed out, ignored for any key insights.

Customer Effort Score (CES) approaches this quite differently — “How hard [or easy] was it for you to sign up for our service, to upgrade to the latest release of our software, to get your home delivery completed on time and accurately if someone had to come into your home to do something?” CES wraps around a very important concept of making it easier for customers to do business with you. That is why we wrote Chapter 4 “You Make it Easy for Me”2 in  Your Customer Rules!: Delivering the Me2B Experiences That Today’s Customers Demand. We discovered that recognized customer experience leaders such as Nordstrom, Vente-Privee, and Yamato Transport (the Me2B Leaders), obsess over making things simple and easy for their customers. Importantly and essentially, they also make it simple and easy for their employees to work with their customers. By making it easy, they know it’s going to win all sorts of positive responses, either in this type of customer effort survey or maybe even in NPS.

There are other exciting customer feedback tools emerging that you might want to explore such as apps on customers’ smartphones. I’ll cover this topic in a future CustomerThink column.

There are other metrics or KPIs that might fit your needs better than NPS, including:

  • CPX = The cost of customer-initiated contacts for certain reason codes divided by the orders shipped, embedded base of products, or some other driver “X”. I’ve introduced this in several earlier columns3 and it’s the basis of our 1st book The Best Service is No Service4.
  • Customer sentiment based on speech analytics of recorded calls or text analytics of chat threads, email, and open-ended verbatim comments.
  • Turnover or attrition of your “best customers”, those with the highest level of future profitability (as opposed to those with RFS, recency/frequency/spend).
  • Social analytics, especially when compared to spend and CPX.

The Future of Customer Experience Metrics

Some of the analysts that follow NPS and Customer Effort Score have shown through various degrees of significant proof that CES leads to significantly higher loyalty than Net Promoter Score. While the jury’s out between these two metrics, whenever we see that a company is only using NPS, we introduce Customer Effort; if they’re only collecting CES, we want to make sure that there’s some connection into a Net Promoter-type of score.

But moreover, we now have at our fingertips the power of advanced analytics with recommendation engines and fast feedback loops, Big Data, smartphone apps, speech and text analytics, and adjacent metrics like CPX. Try them out, and see what works for your company and with your customers. In the spirit of my McKinsey colleague Tom Peters, “If it’s not broken, break it” since your customers’ increasingly high levels of demand will be your best guide.

1 from November 2015.

2 Your Customer Rules! Delivering the Me2B Experiences That Today’s Customers Demand, Bill Price & David Jaffe (Wiley 2015). Based on original research into 12 recognized CX leaders there are the 7 Customer Needs that produce a winning “Me2B” culture, with a total of 39 sub-needs:

  1. “You know me, you remember me”
  2. “You give me choices”
  3. “You make it easy for me”
  4. “You value me”
  5. “You trust me”
  6. “You surprise me with stuff that I can’t imagine”

3 Such as and

4 The Best Service is No Service: Liberating Your Customers From Customer Service, Keep Them Happy, and Control Costs, Bill Price & David Jaffe (Wiley 2008).


  1. Agree with the NPS problems – there are more than what you’ve cited (such as what a company does when its NPS flatlines), but 4 is a good start – but, respectfully disagree that CES is a good alternative ( To take granular action on all aspects of the customer experience, and not just the service element, understanding behavioral drivers is critical. There are excellent examples of where reliance on NPS doesn’t deliver that kind of desired result:


Please enter your comment!
Please enter your name here