This guest blog post was written by Matthew Tworney, Head of Product Marketing, IBM Now Factory, IBM
NPS (Net Promoter Score) as a concept started in 2003. It was developed by Fred Reicheld as part of Bain and Company and Sametrix, and now is a registered trademark. After initial adoption, the main reason why NPS became so important is that it has, in many studies, been directly correlated with business growth. If satisfaction among using services is improved, then revenue grows, which of course makes intuitive sense.
A key difference between Customer Satisfaction (CSAT) indexes and NPS is the way that the questions are phrased. CSAT scores tend to work on questions such as “how do you feel now about the service you just had?” This question is good for judging how satisfaction for services varies over short time periods—think of satisfaction cards in restaurants that people fill out. However, NPS bridges a gap in which NPS looks at how the subscriber feels holistically about the service. This approach is a broader metric to identify those who are happy, those who are thinking of leaving and those who may not do business with you again.
My own personal experience with NPS involved a call from my telecommunications provider to perform an NPS survey, and I happened to be very happy. In addition, the provider has a tariff that offers much better value than competing tariffs from other providers, though it is not always easy to determine which tariffs are the best value. I have convinced ten of my closest family and friends to move to this tariff, and they in turn passed on their good experiences to their friends.
As a promoter, I influenced many other people who in turn went on to influence others. The effects of a positive NPS are huge. The impact from detractors has a similar effect in the other direction. How is NPS measured? The beauty of NPS is that it is one of the simplest scores to measure (see figure).
Based on the simple, “would you recommend…?” type of question, subscribers self-select into one of 11 groups. Those groups are further divided, with the seven lowest groups, comprising 0–6, being detractors and the top two being promoters. The middle two are considered to be on the fence.
The detractors are subtracted from the promoters to determine the NPS. Supposing there are 100 people, for example, the maximum the score can be is +100 and the minimum is –100. NPSs that are in the positive territory are considered good. And a NPS of more than 50 is considered excellent.
Taking this single metric from a sample of subscribers is what drives telecommunications providers across the world, even though it comes from a small variable—a cross section of subscribers. In addition, actions that come out of the survey are difficult to verify because a subscriber may have had a bad experience. But how do we know what element of that experience is causing that subscriber to be upset? Linking how customers feel about the service with how the service actually performed for the customers is not often done.
Looking through the NPS lens
Some inherent limitations of NPS do exist. First, not all customers—such as those in a telecommunications network context—can be asked to rate their services. The actions derived from those surveys can be considered to be small and targeted, with the hope that improving NPS for a subsection of customers can lead to improving NPS for additional customers as a byproduct. In other words, the hope is that when NPS is measured again, the new cross section of subscribers that are asked the questions will have benefited from some of the improvements for those subscribers who were helped previously. Hope in this instance is not a strategy.
Second, NPS data is not brought together with the other rich sources of data within a telecommunication provider. Information such as experience data, customer care calls, churn data, demographics and location awareness are not used to look further into the NPS data to see real correlations between detractors and promoters and the vast amount of stored customer information.
Is NPS for everyone? What is needed is linking NPS to actual customer experiences. And from that approach we can then link all customers to an NPS that is modeled from the original NPS data (see figure). As shown in this chart, combining NPS survey results with many other data sources enables modeling NPS not only for the small section of people who can be asked questions, but for other groupings to get an inferred NPS.
In this way, we can derive an NPS for everyone. If a particular device is causing an NPS issue, then fixing it for all users of the device is important. If a particular location is highlighted in an NPS survey, then it is likely causing problems for many other users in that location.
When an NPS can be inferred for all users, two examples show the view by all locations and devices (see figure). An NPS is determined for the entire network all by taking an NPS from a segment of the subscriber base.
What actions can be taken though this approach? From a number of customer engagements, these real examples (see figure) demonstrate where NPS can affect outcomes and achieve remedies for customers.
In each case, the incidents and the impacts of those incidents may have been lost in the noise of the issues of the day. Where prioritizations are based on outages, alarms and warnings, big impacts to NPS may be ignored because the information needed to make those judgments just doesn’t exist. This new lens provides the ability to look anew at the network for all customers. And we can see where those customers, services, devices and network issues are having the most impact on the NPSs that need to be addressed. In other words, we can use NPS to determine the best way to impact business growth and revenue.