Changes are automatically saved. Reset Settings

Big Foot, the Loch Ness Monster
and Predicting Customer Loyalty—
Which of These is Real?

No Single Number Predicts Loyalty but NPS is the Foundation for Doing it Right


When I moved from branch manager to the marketing department at Grainger circa 1991, I came upon a mystery that market researchers had never been able to solve: how do you predict which customers will be loyal vs. those who won’t? You can tell your customers you love them but for most of them, your relationship is transactional. Which ones are in it for the long term?

The CSAT Stat Was All That


You really can predict which customers are likely to be loyal if you combine NPS with solid customer research.”

Back then, we’d ask customers this very simple question:

Thinking about your last experience with us, were you:

We calculated the CSAT score by awarding 5 points for “Completely Satisfied,” 4 points for “Very Satisfied,” etc. Generally, we’d combine the 5s and 4s and call this group, “Satisfied,” but when we tracked their purchases over time, they were no more loyal than anyone else. What the heck?

Enter NPS

Everything changed in 2003 when Fredrick Reichheld at Bain Consulting published, “The One Number You Need to Grow” in Harvard Business Review, thus introducing the “Net Promoter Score” or NPS. Reichheld reminded us that asking about “satisfaction lacks a consistently demonstratable connection to actual customer behavior and growth,” and instead recommended the question, “How likely is it that you would recommend [company X] to a friend or colleague?”

But Reichheld also wanted to overhaul the scale and calculation that had been failing us for decades. Here’s the scale he came up with instead:

Rating Meaning
9 or 10 Promoters
7 or 8 Passively Loyal
0 to 6 Detractors

And for the calculation, he figured out that you should only count the 9s and 10s as loyal, consider the 1–6 ratings as negatives and discard the 7s and 8s.

Here’s how this system comes together. First, ask the “Macro Indicator Question” using a scale of 1-10 (we’ll get to the qualitative question, “What is the reason for your rating?” later.

Now, add up the percentage of responses that were 9s and 10s, subtract the percentages of 1s through 6s and throw out the 7s and 8s. Here’s an example of a survey that had 121 responses.

  Rating # of
% of Total
Promoters 9 or 10 53 44%
Passives 7 or 8 45 37%
Detractors 1 to 6 23 19%
  Total 121  
NPS Score = 44% - 19%: 25

As you can see, this is a much more sophisticated measurement of customer satisfaction vs. the old method that only accounted for the “4s and 5s” that loved you. NPS punishes you for the customers who do NOT like you very much and so it balances the good with the bad.

The best news about NPS is that it’s now the default method for measuring customer satisfaction across industries; two-thirds of the Fortune 1000 use it “to gauge the quality of their customer experience.” That means you can find a lot of best practices, benchmarks and other useful information to fine-tune your system to your company’s use.

Great—But Does NPS Accurately Predict Which Customers are Loyal?

There’s currently a raging debate regarding NPS’ ability to predict loyalty. If you Google the question, you’ll get conflicting opinions, contradicting research papers and emotional attacks on and defenses of Net Promoter Score’s ability to identify customers who will prove to be loyal to your company. Desperate to find the truth, I turned to artificial intelligence with this straightforward, five-word question:

Does NPS predict customer loyalty?

Instead of a “yes” or “no,” the system generated a 316-word response ending with the kind of weasel-wording worthy of an economist asked to predict the next recession:

In summary, while NPS can provide valuable insights and may correlate with customer loyalty in certain contexts, it is not a definitive predictor of loyalty.

ChatGPT 4.0 has passed the LSAT and MCAT, making it eligible for both law school and medical school, but it cannot give you a straight answer on something simple like whether NPS predicts loyalty. Obviously, ChatGPT plans to go into politics.

Based on our experience with distributors, NPS by itself correlates to purchase volumes and larger customers are more loyal. But to make NPS more predictive, you need to add more data.

The First Step: Qualitative Feedback

Remember, there was an open-ended question in the survey sample above: “2. What is the reason for your rating?” This is where most NPS practitioners—cognizant of the limitations of the “Macro Indicator Question”—hope to get some detail to explain why the customer gave a specific answer.

The good news about this qualitative data is that sometimes customers give detailed answers that really help you dial in your understanding of what is driving their ratings. If you gather a lot of qualitative data and apply “Sentiment Analysis” (another AI tool) to categorize it, you can discern some real lessons and trends about your customer experience performance. Still, this is a pretty imprecise way of understanding what is driving your customer experience ratings—and loyalty.

What’s the Solution to Understanding, Predicting and Improving Loyalty?

Fortunately, there’s a solution to this dilemma. Keeping in mind that what distributors really sell is a great customer experience (product availability has become pretty commoditized), it’s vital that you know what’s driving loyalty with your customers and how you can improve it.

The answer turns out to be NPS + qualitative feedback + customer surveys that gather ratings about various aspects of your services and capabilities. For example, if you measure NPS and it’s a 42, that’s like knowing your temperature is 101.2°. As my business partner says, knowing your temperature isn’t enough, you need some diagnostic work. That’s where the customer surveys are particularly helpful.

The equivalent of a blood panel and an MRI in customer experience research is augmenting the NPS score with more granular and informative data. So, let’s say you’re an electrical distributor and you want to know how to improve your NPS score. Two or three times a year, survey your customers on various aspects of their interactions with you. Common questions will ask them to rate their experiences with you in areas such as:

  • Will-Call Service
  • Delivery Performance
  • Telephone Service
  • Sales Rep Effectiveness
  • Product Assortment
  • Product Availability
  • Pricing
  • Technical Knowledge

Of course, you can add in questions about your services—maybe you build panels, put together cable assemblies, do kitting and labeling, etc. Just be sure the survey doesn’t go longer than 5-7 minutes or you’ll see significant falloff in the completion rate. To help keep the survey length down, you can ask the same questions about core touchpoints every quarter but rotate different questions about value-added services. If you carefully gather information about customer attributes as well (location, company type, title, etc.) you can understand how your performance varies by segment, job function, geography and other factors—and it will vary.

Matching the customer survey with the NPS score will tell you not only your “temperature,” but how you can improve the health of your company. If you have a statistician on staff, you can calculate correlations between various performance metrics and your NPS scores.

If you want to consider an all-in-one packaged solution that ties all of this together, reach out to my company, Distribution Strategy Group, and ask about our new NPS and diagnostic system for distributors called, “Customer RX.”

Operationalizing NPS

Even more important than the customer research is what you do with it. Once you understand the drivers of your NPS score, it’s essential that you build initiatives designed to improve your customer experience over time. Since most distribution leaders are good operators, we often find them adept at taking in this data and acting on it effectively. Some distributors use formal operating systems like EOS to run their businesses and they are often particularly good at driving improvements in customer experience—once they have the information they need to know what to do!

Sorry, Loch Ness Monster and Big Foot—Predicting Customer Loyalty is Real

Market research has come a long way since my early days in the Grainger marketing department. You really can predict which customers are likely to be loyal if you combine NPS with solid customer research.

Just as important, you can use the same approach to understand which customers are not likely to be loyal—and do something about it before they defect. Distributors generally don’t measure the value of customer churn; it’s a huge blind spot that costs them dearly in lost revenues that could have been retained with relatively little effort.

Identifying loyal and non-loyal customers used to be akin to trying to spot Yeti or Nessie, but that’s not true anymore. The methodology I’ve laid out here will help you improve your customer experience, which will help improve customer loyalty. That’s key to driving growth efficiently in your company—and for the long term.