The main customer experience measurement and improvement systems – Part 4 – The Net Promoter System®

Fred Reichheld wrote his landmark article The One Number You Need to Grow and published it in the Harvard Business Review in December 2003. He had spent two years investigating customer loyalty research. What he found was surprising and counterintuitive, at least at the time. Answers to a single question predicted customer loyalty and therefore revenue trends in most industries. A further finding was that “Our research indicates that satisfaction lacks a consistently demonstrable connection to actual customer behavior and growth.” He pointed out that the American Consumer Satisfaction Index for Kmart, for example, continued to improve as it slid into bankruptcy.

Fred and his team studied 4,000 customers across 14 industries, using the 20 questions on the Loyalty Acid Test questionnaire he had designed with his Bain colleagues. The study was about the actual behavior of people; whether they actually bought more and whether they actually recommended a product or service to someone they knew. In 11 of the 14 industries, a single question had the greatest statistical correlation: “How likely are you to recommend X to a colleague or friend?” In two of the remaining three industries, the recommendation question was just behind the top question. This surprised Fred, who had expected asking customers to what extent they agreed that “This company has earned my loyalty” to be the top-performing question.

Rating scale and behaviors

Even in the HBR paper mentioned earlier, Fred was able to do at least two further things. The first was to design a rating scale that was unambiguous. The second was to observe and categorize the behaviors of people who gave different ratings on his zero-to-ten scale. He found that people who give a 9 or 10 tended to buy more and to actively promote the product, service or brand. Those who gave a 7 or 8 were “passively satisfied” and tended not to say much. The remainder tended to speak and behave negatively, and he called this group Detractors. The Net Promoter Score became the percentage of Promoters, less the percentage of Detractors. Fred initially expressed this ‘Net Promoter Score’ as a percentage, and later dropped the percentage sign.

Partnership with Satmetrix

Fred worked with Satmetrix who began tracking 10 to 15 thousand recommendation responses from about 400 different companies in early 2001. In the airline industry, for example, the Net Promoter Score correlated directly and unambiguously with growth from 2000 to 2002. There were industries without such a correlation, mainly effective monopolies, such as cable television companies.

Evolution to a system

The HBR article evolved into the Net Promoter System, a Service Mark owned by Fred Reichheld, Bain and Satmetrix. Satmetrix also provides software and training courses that lead to certification as an NPS professional. The entire system was initially described in The Ultimate Question, now in version 2.0 and co-authored by Fred and Rob Markey, also of Bain. There have been a number of evolutions since the latest book was published and these have been described in some episodes of the excellent Net Promoter System Podcast. You can find the podcasts and many other resources at The diagram at the top of this article illustrates the overall system.

Current state of the Net Promoter System

Reichheld and Markey have made extensive updates to the system that have not yet been documented in a book. The first and most important addition is a second open text question. The result is a standard survey format for most purposes that looks like the illustration below.

NP survey

Episode / Transaction feedback

For surveys about an interaction that has just taken place, Reichheld and Markey have added an additional rating question, “To what extent has your latest interaction changed your willingness to recommend?” just after the first recommendation question. This is measured using a text-based five-point scale, ‘Increased a lot’, ‘Increased a little’, ‘Not change’, ‘Decreased a little’, ‘Decreased a lot’. Rob Markey refers to this as “The Bain four-question progression for episode feedback.”

The concept of an ‘episode’ is an important one. Customers interact with companies to get certain things done, for example to get make a claim on their property insurance after a flood. For the customer, that is an ‘episode’ in the overall ‘TV series’ of their relationship with the insurance company. The insurance company breaks a claim process down into a number of transactions, such as a call to a service center. In this example, Bain recommends getting customer feedback only about the overall insurance claim process, and not about any individual component, such as the call to the service center.

Main features

Other than the metric, the main features of the system as described in the book and in practice are:

  • It is by far the most common customer experience measurement and improvement system. About two-thirds of medium to large companies use it, at least as a measurement and reporting system. Implementing all elements of the full system takes a lot of time and effort, and few companies have a complete implementation.
  • When implemented properly, there are only a few questions, so response rates are higher than in other systems with more questions.
  • The Inner and Outer Loops for improving things for your customers and internally are described well and are easy to follow.
  • There are a number of sources of double-blind competitive benchmarks, making it reasonably easy to understand how you compare to others in your industry.
  • Because it is easy to communicate, it is relatively easy to implement the basic measurement and reporting systems.
  • If you need automation, there are loads of vendors who can help.
  • The credibility of the system depends on your ability to establish the relationship between metric trends and market share trends. This is difficult to do, particularly if you have no measurement history. People have to trust you until there is enough data. That can take a couple of years, and you could lose sponsorship by then. Of course this would be the same with any other measurement system that claims to predict revenue.
  • The main challenge is the analysis of the verbatim responses. Text analysis software should be used for initial screening and categorization to avoid human bias.
  • Finally, because the system is simple, it is also quite easy to manipulate.

Personal opinions

I have worked extensively on the implementation of the Net Promoter System at HP and have provided coaching to others since then. If I had to pick a single thing as its most positive aspect, I would pick communication. It is far easier to communicate how the scoring and system works than with any other system I know. My second observation is that who you ask matters. If you do not ask the correct people, the results will not predict your revenue correctly.


Lots has been written about the Net Promoter System and tens of thousands of companies have implemented at least the score. When I say lots of people have written about it, I have to include the book I wrote and which my brother illustrated: Net Promoter – Implement the System. As always, your views are welcome below.

Next time in part 5 of the series: The Wallet Allocation Rule.