Net Promoter Score: Why most companies are doing it wrong

July 15, 2015
Reading time:

Why companies do NPS wrong

Last month we wrote about how Slack uses NPS to fuel its rocketing growth. These are the guys who know their NPS.

Slack are not unique: Linkedin, Uber and the high flying currency exchange startup TransferWise, use NPS as one of their key health indicators. In fact, we spoke to dozens of successful startups in the UK and US and found that many of them are either already using the NPS system or are considering it.

It makes sense — NPS is probably the best way to measure brand loyalty. When done right, it can be the actionable metric for enhancing your product experience to deliver delight. There is a problem however — a lot of companies are doing NPS wrong. Too often the survey is administered at a bad time to the wrong audience in a non-engaging manner. The results of such surveys can be at best meaningless and at worst misleading. Even when the NPS survey itself works well, companies often struggle to understand why their customers gave the recommendation rating they did.

Here are three most common NPS mistakes:

The Survey — No One Likes It

Survey Respondents

There are several ways of delivering an NPS survey but the most common one is online surveys. The company sends an email with a simple survey link to its customer base. This is the first problem — clicking on the link requires work and people have little visibility on what is waiting for them on the other end. As a result the response rate suffers and the company gets a far from complete picture of how customers are feeling.

Few of those who do click through reach the end of the survey. This is because, quite frankly, most of the time the survey sucks. The Net Promoter Score question is buried among endless other questions — people just don’t have the time. We found that companies who ask short n’ simple surveys get far better response rates and quality of answers. Usually NPS comes at the end of the survey and partial responses are not really recorded in any way. The power of the NPS comes largely from its simplicity so it does not always mix well.

Solution: quick, well designed NPS-only surveys. There are a lot of good NPS templates online. We are planning to release our own to the public soon. Typeform provides the best experience for the survey itself.

Sampling — Too Random

Random NPS Sampling Bias

The only way NPS can be useful is if you can trust it. Getting an NPS you can trust is hard. You need capture a representative sample of your customers. The best way of doing so is to collect NPS continuously and report a representative average over one month/quarter/year. The problem with this is sending NPS surveys on a continuous basis and pulling together the data without an automated setup is a pain.

We have seen many examples of NPS collected once or twice a year from a very small subset of customers being treated as the “truth”. In such situations NPS can fluctuate by over 30 points between and cause more confusion than insight. Imagine an e-commerce startup measuring its NPS only once, after a Black Friday sale when the customers are guaranteed to be happy. Random positive as well as negative events can affect the score too much and lead the company to focus on issues outside of its control.

Solution: The best way to run an NPS is to send a survey to each customer X days after a certain touchpoint such as first purchase or list subscription. It’s good to add a bit of variation.

“We ask every customer the NPS question. However, we don’t ask them all at the same time. (e.g. Some get asked soon after their 1st transfer, some after their 2nd transfer).” Nilan Peiris, VP of Growth, Transferwise

Analysis — More Questions Than Answers

Probably the biggest headache with NPS is doing something useful with all the data once you have collected it. Calculating the score is easy. However that is not where the real value of NPS is. Business results come from a robust understanding of why your customers would / would not recommend you. Unfortunately, this involves reading through and tagging thousands of customer comments. Most companies, especially high growth startups, do not have the resource to do this right. Thus the insight never gets unlocked.

Solution: Use a tool like Chattermill to automatically categorise all comments. Alternatively, do it yourself. Spreading out the survey over time will help a lot.

The most actionable part of the NPS survey is the categorization of the open-ended verbatim comments from promoters & detractors. Each survey we would analyze the promoter comments and categorize each comment into primary promoter benefit categories as well as similarly categorize each detractor comment into primary detractor issue categories. Rachin Rekhi, Director of Product Management, Linkedin

A Better Solution

We admit — the above is a daunting task. No one wants to read through thousands of comments. If only there was a way to automate this process…

There is.

At Chattermill we have developed cutting edge Natural Language Processing technology geared specifically towards analysis of NPS comments. We can automatically classify your free form text into granular categories and even identify issues that have the biggest positive/negative impact on customer loyalty.

Customer Theme and Sentiment  Data Analysis

Feel free to get in touch for more tips on how to make the most of NPS.

See Chattermill in action

Understand the voice of your customers in realtime with Customer Feedback Analytics from Chattermill.