A lot of marketers and executives love NPS®. The appeal is clear – a single number that can tell you how happy your customers are and, as its creators claim, can actually predict the finical success of your company. All this from one single question.
The NPS score looks good on a dashboard and it provides the coveted “single number.” There’s a problem, though.
In and of itself, your NPS score is totally useless.
Like the scores on standardized tests you took in school, the score doesn’t mean much by itself. What mattered about your test scores was how you did relative to everyone else. Similarly, your company’s NPS score can only be interpreted in the context of how other companies perform and how your company has performed in the past.
Even then, the score might tell you that things are “good” or “bad”, but don’t give you any actionable information that you can use to improve your company’s performance.
So how can you make the NPS tracking that you’ve invested in useful?
The NPS score is the start, not the end, of understanding your customer’s satisfaction. Only by understanding the context in which a customer scored your company the way they did can you get actionable information.
How can you do that?
Find out why
The first step is to follow up the NPS rating with an open ended question that asks why. This is recommended in the Reichheld book, but I’ve seen many cases where this was omitted by people implementing NPS. This step alone can greatly increase the value you can get from NPS, but there’s more that can be done.
Applying quantitative analysis techniques to the NPS score and the reasons given for the score can uncover patterns that help you understand what elements drive satisfaction. The easy way to do this is to ask a series of rating questions about various aspects of the customer experience and to perform a regression analysis on the data. The danger in this approach is that you miss what actually matters to customers.
An alternative is to code the open ended responses and use those codes as explanatory variables in a quantitative analysis. Depending on the volume of responses you are working with (I’ve seen companies with tens of thousands), a manual coding of open ended remarks may be impractical or prohibitively expensive. Manual coding of large volumes of open ends also invites data problems from inconsistent coding of responses.
Some of the current text analysis software does a good job of automating much of this process (no thoughtful analysis can ever be totally automated), which makes the processing of large datasets much more feasible. At a minimum, these systems speed up the process and ensure consistency in the coding of comments. The best systems can also help interpret meaning and sentiment, as well as uncover relationships (X and Y are often mentioned together, for example).
Once coded and used in quantitative analytics, a much better picture of customer satisfaction starts to emerge. As you start to understand why your NPS score was what it was, you can take action to improve business outcomes.
Round out the data with who and where
By adding some profile and purchase context information to the mix, you can begin to understand what customer groups you are serving more or less successfully. This information helps you find other people like the happy customers, focusing your customer acquisition activities.
So, NPS is a good start, but it isn’t enough to really understand your market or successfully compete. Add the understanding of why (open ends), who (profile), and when (purchase context), apply some thoughtful analysis, and you’ve got a recipe for real customer insight.
If you’d like a copy of this article suitable for printing, click here for a PDF version.
® Net Promoter, NPS, and Net Promoter Score are trademarks of Bain & Company, Inc., Satmetrix Systems, Inc. and Fred Reichheld