This article orginally appeared on LinkedIn.

My last blog on the importance of integrating survey NPS^® with signal NPS, NPS: It’s Not about the Score, generated a lot of conversation and questions. While it’s true that I coined the term “signal NPS” within the past year, the concept isn’t new. The idea of a signal Net Promoter Score^® actually preceded survey NPS.

During the 1980s, Bain & Company teams identified the power of loyalty economics. We focused on actual customer behaviors such as repurchase (or retention) and increases in average ticket and share of wallet. These observable behaviors clearly signaled that a customer was happy, and that the firm had created an asset: a loyal customer. 

One problem with focusing solely on signals was that they often lagged the triggering events that generated loyalty or disloyalty. A bad claims experience might happen in June, but the auto policy renewal decision didn’t take place until December. The claims rep who handled that customer had probably transferred to another job by then, maybe even gotten promoted. The lag time was limiting what we could learn, and making it hard to hold employees accountable.

When a customer defected at that anniversary, we encouraged our clients to call that customer and probe for the root cause for their decision to move to another insurer. I wrote about this process in several journals, including a piece in the Harvard Business Review called “Zero Defections: Quality Comes to Services.” The management process we developed identified the key signals of unhappiness—defection and reduced purchases among them—and triggered outbound calls to learn what went wrong and how best to fix the problem. Even if it wasn’t possible to win back that customer, the organization could reduce or eliminate the issue before it affected other customers.

One large insurer studied what triggered defections by coupling analytics with the root-cause interviews of defectors. The company looked at a large sample of customers who failed to renew their policies and then identified what transactions they had with the insurer during the 90 days before defecting. Did they have a claim? Did they call the service center? Did they change their coverages? And so forth.

Next, it identified the transactions that happened most frequently for defectors and compared them with customers of similar tenure, age and risk who renewed. This enabled the company to pinpoint the transaction most closely associated with defection: a customer change of address. Yes, when customers move, it turns out, they reevaluated their policy choices. Moving was the signal that risk of defection was highest. 

Knowing this, the company could experiment with various ways of delighting customers when they reached this junction. For example, it negotiated prices with highly rated moving companies, then offered the lower prices to its customers. The insurer also gave incentives for the current agent to retain the business by introducing customers to the company’s agent in the town where they were moving. There were many innovative ideas. The company piloted the most promising and tracked whether or not those actions improved the retention rate among clients who moved. 

Today, of course, there are sophisticated big data tools that might make the manual labor involved decades ago obsolete, but it remains important not to overcomplicate this process. Keep things practical by identifying the one or two most important signals—and then test whether innovations and actions result in improved loyalty as measured by both survey NPS and signal NPS.

Over the years, the core concepts haven’t changed. We simply have better tools and more precise categories that can accelerate learning and innovation. In this way, we’ve arrived back to the future.

Net Promoter^®, Net Promoter System^®, Net Promoter Score^® and NPS^® are registered trademarks of Bain & Company, Inc., Fred Reichheld and Satmetrix Systems, Inc.