Why the creator of NPS thinks you should look at customer verbatims
by Mikhail Dubov
Net Promoter Score has taken off like few measures have in recent years. Companies from Apple and Tesla to HSBC and Lego are using it and even though the metric is specifically designed to be simple, people still argue what is the best way to run an NPS program.
One thing almost everyone would agree on is that just measuring NPS is not enough. If you turn it into a “vanity metric”, you might as well drop it completely. Recently, Fred Reichfeld, the brain behind NPS, recorded a great podcast where in a wide ranging discussion he touched upon what I would argue is the most important part of the NPS, analysis of customer feedback.
Don’t over-complicate things
NPS was developed to find the perfect balance between a metric that has high explanatory and predictive power across different industries and types of businesses and a simplicity which ensures it is actually used throughout a company. The beauty of NPS (and similar metrics such as CSAT or Customer Effort Score) is that you can communicate them across your organisation from investors to customer support agents.
The original search was to see if there was one question that would get the lion’s share of the insight and it turns out it has about 70–90% of predictive power. … Could we come up with a model of 50 questions that is more accurate? Of course! But what you want is a model that drives learning and action, not a score that has some PhD thesis attached to it. — Fred Reichheld
Always attach a verbatim explanation
As already hinted above, NPS is not about a number but about learning and improving. No matter how robust your score is, the really important thing is understanding the drivers and using this knowledge to improve experience.
The best way of learning is to ask customers to tell you what is important in their own words. Removing the burden of tedious multiple choice questions, you find an incredible wealth of insight. More importantly, it’s the insight you have not already thought of. You might think yourself clever asking a question such as “Rate your delivery experience”. In many settings, such as e-commerce, delivery is indeed the key problem. But asking that question is actually counter-productive. People will answer it with a 1 or a 5, but you will never learn if delivery was actually the main thing that sticks with them or the fundamental reason why they would or wouldn’t use or recommend your service.
There is no way to get the richness, the colour, the detail from the multiple choice question that you can from a conversation. The little things contribute to your impression of the product and no multiple choice question will tease this out.— Rob Markey
Every customers feedback is important
Quite often we come across companies that only look at the detractors, i.e. those answering 0 to 6 on the “Would you recommend company X to a friend?”. This is far better than not listening to any customers (which a lot of companies unfortunately also choose) but this misses some of the most important comments that come from your Passives and Promoters which often also contain negative comments flagging areas they are unhappy with.. Consider the following comment that we found online:
The customer is obviously happy. She gave a score of 9 and even reading the comment you can see she is definitely a promoter. But our algorithm also identified a negative mention of size. True, it did not sway her opinion, but it might be a problem for someone else. So you can learn not to offer this product to customers for whom size is important or that size is in fact important to customers such as as this one. If you only looked at the detractors, that insights would be missed. Moreover, if you miss little things promoters aren’t happy with, you will eventually start loosing your most loyal customers.
How to cope with volumes
When Reichheld and Co were writing their first book, it was inconceivable for a large company to ask every customer for feedback and to read each one. If you try to do it yourself, you will fail as soon as your volumes get into 100-s. As a stop gap measure the creators of NPS suggest going through at least a sample of the comments.
Fortunately, modern technology allows you to break the trade off between quality and quantity of insight. With deep learning, we can understand every single comment and provide you with an aggregated view of what drives your feedback.
This post originally appeared on Chattermill blog.