​This article originally appeared on LinkedIn.

Most companies don’t really know who their best customers are. And that makes it hard to get the most out of a customer-centric philosophy.

So, who are your best customers?

Are your best customers the ones who generate the most revenue? Some leaders believe that their biggest customers are the best customers, but those large customers may not represent a growth opportunity. Your firm may have already saturated their needs. Also, size can bring challenges. Big customers may throw their weight around and demand concessions on pricing and terms. Because of their size, they can threaten, cajole, and demand product features, order quantities, and special features that increase your costs. Knowing how important they are, big customers may bully or even abuse your employees. Some customers just don’t believe in being good partners. Loyalty is simply not part of their philosophy. These are some important reasons why biggest does not always mean best when it comes to valuing customers.

Are your best customers the ones who generate the most profit? Perhaps these less-demanding customers don’t soak up extra costs or services. They don’t, for example, return products even if they aren’t very happy with them. In fact, you may not learn about their disappointment until they defect to a competitor. Price-insensitive customers may make it easier for you to earn a profit today, but don’t you need demanding customers to keep you on your toes and force you to innovate, improve, and be more efficient? As long as those tough customers are demanding services and features that lots of other customers will value, responding to them can boost overall profits and success—even if these individual customers are less profitable individually.

Are your best customers the ones with the highest lifetime value (or net present value), as I’ve heard some experts argue? That seems to make a lot of theoretical sense, but how many of your team members understand what a net present value is, let alone how to calculate it? They would need to be trained in finance and actuarial science. And even if these sophisticated measures are developed reasonably, often too much of the expected value depends on the probability of future events and decisions that won’t play out for many years. With this degree of uncertainty, how can your team spot these customers and treat them accordingly? And how can investors really trust these conceptual estimates?

If you’ve read Winning On Purpose, you already know that I believe your best customers are the ones who will enthusiastically and genuinely recommend your business to a friend or family member. Cultivating such promoters is the only lasting way to build a sustainable business.

Unfortunately, too many firms are so obsessed with achieving a high Net Promoter Score that they have come to believe that the best customers are the ones who consistently provide 10s on surveys—and post positive online reviews. But so many reviews have been “bought” or manipulated through quid pro quo that their real value seems to be declining. And customers who provide courtesy 10s provide little guidance for required improvements you need to remain competitive. Frontline teams are not inspired when 10s no longer represent a standing ovation but merely confirm that nothing really bad happened. The goal should be to generate word-of-mouth referrals from happy customers.

And with new technology systems, these referrals can now be tracked and understood. One such system developed by London-based software firm Mention Me—a company I am an investor in and adviser to—is beginning to reveal remarkable insights about customer referrals, including some big surprises about which customers really generate the most value. Mention Me tracks not only when customers actually make referrals but also the subsequent series of purchases and second-generation referrals generated.

Mention Me CEO Andy Cockburn explained to Rob Markey in a recent episode of the Net Promoter System Podcast that his system identified a customer who spent only $261 at a luxury retailer but kicked off a network of referrals and subsequent purchases that generated $530,000. Mention Me’s analysis of its scientific database of advocacy behavior and economics shows that the true lifetime value of most customers is dominated by referral economics, not by that individual customer’s lifetime purchases. Remarkably, the volume of an individual customer’s purchases has no relationship to their propensity to refer.

I suspect that as more and more companies use technology to track referrals, leaders will conclude that the most important (and best) customers are indeed the ones who are so happy that they refer others. Understanding the root causes of their behavior (and optimizing this advocacy engine) will naturally become the primary focus for learning, innovation, and improvement.

Of course, once you decide which customers are most important, you must share this knowledge beyond the headquarters staff. Customer service, product and experience design, shipping, and production scheduling will all need to understand how to gauge customer value to ensure the best customers receive appropriate white-glove treatment and are the design target for next-generation products and services.