A guide to building more insightful NPS surveys

NPS — or Net Promoter Score — is one of the simplest and most widely used metrics that companies utilize to understand how they are performing in the eyes of a customer. It uses a simple question that captures the essence of the fundamental purpose of any business — serving its customers: “How likely would you be to recommend us to your family or a friend?” Typically, this question is asked on a scale from 0–10, with 0 being “Not at all” likely to recommend and 10 being “Extremely likely” to recommend. The NPS score is then calculated as a difference between promoters (% of respondents scoring 9 or 10) and detractors (% of respondents who scored 6 or below).

This concept is beautiful, because it can be boiled down to one simple question that can be asked quickly and unobtrusively. In return, it provides a good benchmark of a company’s performance over time and against competitors. However — and I bet anyone who uses this metric regularly would agree — as the NPS score becomes an integral part of managing a business, it ultimately leads to an even more fundamental question for the management: “Why is our NPS score X? How do we make it to increase to Y?”

Fortunately, the “why” behind the NPS score is discoverable with a good survey design (Note: going forward, I use “survey” and “study” interchangeably in this guide), so without further introduction, let me share with you what has worked for me and for GroupSolver’s clients.

Keeping it simple

The beauty of the NPS score is that it is simple and thus it is an antithesis to a typical customer satisfaction survey, which can go on forever. While understanding the “why” behind the score requires adding a few more questions to the study, we still want to keep them to the bare minimum to avoid upsetting the customer into quitting the study before answering the last question. As is good design practice, I always advise our clients to think about every question in a study and its purpose. If the answer has a potential to fundamentally change what we do tomorrow, let’s ask it. If it is simply a “nice to have” or “could provide interesting insights”, we cut it out.

In the context of an NPS study, there are usually only a few questions that really matter:

  • How likely would you be to recommend us to your family or friend? (the core NPS question)
  • What must we improve for you to recommend us? (the “why” behind a low NPS rating telling us what we need to do differently)
  • What are we doing that would make you recommend us? (the “why” behind a high NPS rating telling us what we should be doing more of)

Sometimes it makes sense to ask a few additional questions to understand the customer who is providing the recommendation. This helps businesses better understand which customers — or which shopping situations — yield better customer satisfaction. This is, however, an area where it is easy to make the mistake of asking 20 extra “nice to have” questions, making a study too long. If we need this additional information, what often works well is asking questions such as:

  • Why did you come into our store today?
  • What are the most important things that matter to you when you shop for this product?

Using open-ended questions here helps reduce the load on the respondent. an additional benefit of using open ends is that such questions allow our customers to tell us what is on the top of their minds rather than just validate hypotheses we came up with for them.

An example

Rather than talking theoretically, allow me to share a specific example. Let’s build a simple NPS study that will measure your experience with reading this guide.

Before I ask you the first question, I want to tell you that this will only take 2–3 minutes of your time. I like to be transparent with what I am about to ask you to do for me, and I want to let you know that I respect your time.

Next, let’s ask the core NPS question. There is no need to ask anyt question prior to the NPS question. The best practice is to ask the NPS question as close to the actual experience as possible. We want the response to be coming from the “gut” — recommendations are usually emotional in nature — and we want respondents to experience as little priming as possible between the experience we are measuring and the question they are answering.

Notice that we provided a description to both ends of the scale. In scales that are shorter, I always provide descriptions to all the points on the scale. But in this case, it would be impractical. NPS questions are quite standardized and it is best to stick to the standard format if I want to compare my NPS score to other guides out there.

The next step will take us to the most interesting question I want to ask you — “why did you (or did you NOT) find this guide worth reading?”. We have a choice to make here: do we ask respondents about both things I should improve AND things I am doing well? Or do I just ask you one question? Speaking for myself, I rarely have only positive or only negative things to say about a product or an experience, so my gut is telling me that I want to hear the good things you have to say to me in addition to your criticism.

However, I recognize that the factors (themes) — positive or negative — are likely going to be consistent for those who generally liked this guide with those who had a mixed experience. So, I choose to only ask you about the positives OR negatives depending on what score you gave me. If you scored this guide higher than 7, I will ask you about what you liked and if you scored 7 or lower, I will ask you about things I must improve. This will make the study shorter and I will still hear the key answers and themes to help me improve this guide.

I could finish the study right here, but in the future, I may want to make different versions of this guide specifically for students or for readers who want to learn more about how the GroupSolver platform works. Therefore, I will ask additional two questions:

Let me point your attention to two things. First, I chose to ask about your reasons for reading this guide as an open-ended question. I may hypothesize why you may be reading this, but I actually want to hear this in your own words. Why? Because I want to understand how you talk and think about the reasons for being here, so that I can target my outreach better in the future. Furthermore, chances are that you will provide an answer or two I would not have considered by myself if I were to make this a traditional multiple-choice question.

Second, in contrast to the previous point, the question about “who you are” is much more hypothesis driven, which is why I used a multiple-choice question. I wanted to reach out to a few specific types of audiences, and, in the future, I may want to build a guide tailored to their specific level of interest and background knowledge. Therefore, I want to segment the NPS scores I receive and the reasons for them by those demographics. This will allow me to effectively build and position different versions of my guide.

On a final (and quite obvious) note, I need to mention that keeping this study simple and still getting to the insightful “why” behind the score necessitates the use of open-ended questions.

Every survey tool gives you the ability to ask open-ended questions and I strongly encourage you to use them. With the advances in text analytics and machine learning, researchers don’t have to be afraid of using open-ended questions. Advanced customer insights platforms — such as ours at GroupSolver — can help you convert open-ended questions into quantified data without the need for manual coding, which is time consuming and potentially biasing.

Thank you for reading, and if you have a couple of minutes, please do give me your feedback: Customer satisfaction study