Can a single survey question serve as a useful predictor of growth for a company?
Yes, yes it can. How likely is it that you would recommend our company/product/service to a friend or colleague?
That question is used to calculate a company’s Net Promoter Score (NPS). The NPS question is answered on a 0-to-10 Likert-type scale (remember high school?), with 10 being “definitely likely”. “Promoters” are those who give ratings of nine or ten to the question. The “passively satisfied” are a seven or an eight. And the “detractors” score from zero to six.
You calculate NPS by taking the % of promoters (number of respondents who ranked 9 or 10, divided by total number of respondents) minus % of detractors (percentage of those who chose ≤ 6). It essentially only counts your mom’s and your ex-girlfriend’s ratings of you. And, it cuts out the high school acquaintances you occasionally see on the street and smile at, but hope they never actually strike up a conversation.
It took Frederick Reichheld two years of dedicated research to zero in on this one question and how to score it, which he published in the HBS Review in 2003. It is a long read, but it is one of the most interesting pre-YouTube (launched in 2005) pieces of content I have seen on the internet.
By limiting the promoter designation to only the most enthusiastic customers, not just those who choose above a 5, the scale avoids the “grade inflation” that often infects traditional customer-satisfaction assessments and Economics 101 classes across the U.S., where fresh baked cookies and a call from an eager mother can throw off the curve.
Customers just above neutral are considered satisfied, but those customers don’t help your score. That’s not enough. The idea is that the only path to profitable growth lies in a company’s ability to get its loyal customers to become, in effect, its marketing department.
Take Tesla, who consistently tops the leader board with an NPS of 96. I got a text from a friend who test drove a Tesla and said it was the most incredible experience he’s had. I said Watt, you haven’t even purchased the product yet, and you are already helping the company sell? It’s game over once they make the purchase. And because of that, Tesla doesn’t have to advertise.
Comcast sits at the bottom, with an NPS of -5. That’s not a typo. That is a negative 5. If you are shocked, give Comcast customer service a call today and email me tonight when / if you get off the line. Amazon has an NPS of 62, and in light of the recent HQ2 debacle, Amazon has stopped reporting its NPS score. They’ll probably wait until after Prime Day to report it again. Nike is at 32, and after arguably the most famous shoe incident in college basketball history, I expect that number to continue to drop quite quickly.
Let’s take a step back. What is the goal of customer surveys? You want to measure your customer’s satisfaction and dissatisfaction in a cost-effective way and then use that data to improve your growth and retention.
What Reichheld did was substitute a single question for the complex black box of typical customer satisfaction surveys. Those surveys are complicated, yield low response rates and provide ambiguous implications that are difficult for operating managers to act on.
The Net Promoter Score works, because it allows you to keep it simple. Too many of today’s satisfaction survey processes yield complex information that’s months out of date by the time it reaches the managers on the front line. The NPS number and its underlying data gets you to action more quickly. A customer feedback program should be viewed as an operating management tool, not as pure market research.
The question itself works because it shows how many customers, and which ones, are putting their reputation on the line for your product. Granted, doing so is more valuable to some than others, like Peyton Manning endorsing Papa Johns. Come on Peyton, there’s no way you enjoy that.
Reichheld chose this exact question because traditional customer satisfaction statistics lack a consistently demonstrable connection with company growth. In some cases, there is an inverse relationship. For example, K-mart experienced a significant increase in the company’s American Consumer Satisfaction Index in the early 2000’s, which was accompanied by a sharp decrease in sales as it slid into bankruptcy. I promise that will be the most out of date statistic I'll use in this blog.
To get to this wording, Reichheld tested out 14 different questions. He tied survey responses from individual customers to their actual behavior, including repeat purchases and referral patterns over time. “How likely is it that you would recommend Company X to a friend or colleague?” blew away all other questions in terms of statistical correlation with repeat purchases and referrals. The second and third contenders were “How strongly do you agree that Company X deserves your loyalty?” and “How satisfied are you with Company X’s overall performance?”
The standard NPS question doesn’t always work for all industries. In database software companies, for example, senior executives select the vendors. So, it doesn’t make sense to ask users of the system whether they would recommend the system to a friend or colleague, as they had no choice in the matter. For those industries, try using phrases that understand if the Company “sets the standard of excellence” or “deserves your loyalty”.
It would be nice to expand the universe of what NPS was used for… How’s your son? He’s like a 55. It’s not that I wouldn’t recommend him. But, something happened at school last week. And you know, we’re just cautiously optimistic for the college application season.
When analyzing a company’s NPS, obviously the higher the number the better. It is important to look at a company’s NPS relative to its competition. NPS and its trends are a great leading indicator that the company is not only focused on its users, but is improving its value proposition over time.
As an investor, look out for funky ways of showing NPS. Like only showing percentages of promoters, not including detractors. Also look into how they are showing the number choices, especially on mobile. As the layout of the number choices has led to scores that can differ by 20 points depending on if you put 10 at the top and scroll down to 0, or vice versa. Experts (yes there are multiple experts on a one-question survey) recommend doing a 50/50 split on phone screens.
I’ll take this time to congratulate the ComedySeller for its impressive NPS of 95! Thanks to all eight of you who filled out our real NPS survey. I know it was anonymous, but I took the liberty of unsubscribing Aunt Carol, who brought the average down, and coincidentally is no longer known as the “cool Aunt”. Good luck at Thanksgiving Aunt Carol. Can’t wait to review your turkey.
If you liked this post, recommend it to a friend. If you didn’t, let me know here.
I am a 25 year-old venture capitalist and amateur stand-up comedian living in NYC.