What you need to know

• You can't measure your weight to fitness, says Rob Markey, a pioneer in the NPS world and the founder of Bain’s Customer Strategy & Marketing practice.
• Instead, he urges organisations to develop a Net Promoter System with three levels (he calls them loops).
• The first involves individual employees collection individual feedback in real time, one piece at a time.
• Next, get teams of people in similar jobs together to talk about what they have learned.
• And finally, issues that can't be solved by the teams are routed up the chain to the outer layer for consideration and action, and importantly, accountability.
• And none of this works without technology, Markey says.

Just like at the gym, you can't measure your way to being stronger, says Rob Markey, Bain Advisor Partner and founder of the company's Customer Strategy & Marketing practice. "You don't buy a better tape measure to get stronger, you actually have to work out at the gym."

In an interview with Mi3 Australia at last week's Qualtrics X4 conference in Salt Lake City where he was speaking, Markey said the concept of the Net Promoter System evolved from the observation that the radical simplicity of the score was great, but not sufficient.

“We figured out there were some systematic things consistent across the companies using it well versus those that were using it poorly," he said.

Markey described what he called the three basic elements of the Net Promoter System. The first is the inner loop of the Net Promoter System where feedback is collected from individual customers by individual employees in real-time, one piece of feedback at a time.

"If the feedback merits follow up with the customer, then you follow up with that customer, partly to reward them for having given you feedback, partly to repair relationships when needed, and partly to learn more, even when the customers had a really good experience with you. If they share a lot of feedback, they're telling you they have more to say," he explained.

Also, Bain realised by giving that feedback to frontline employees immediately, those employees learned much faster. 

"The traditional way of handling customer feedback was to aggregate it into reports and deliver it in big chunks, once you have amassed statistically significant results. You can still do that with Net Promoter, but the most valuable thing is: 'I do something - there's an interaction or reaction - if I do it in a different way and there's a different reaction.’  That learning cycle is enabled by this. That's the inner loop.”

The middle loop involves getting teams of people with similar jobs together to talk about what they have learned. It’s a technique Bain developed working with Telstra, said Markey.

“Huddles were developed to get people to get together regularly and be engaged with customer feedback. That lead them to ask  ‘How are we doing? What is working, what’s not working? What can we individually do to make things better?" he said.

Just as importantly, it also identified where the staff in organisations need help when it relates to the things they can’t control such as pricing policy, product or technology.

“Those issues are then elevated into what we call the outer loop where they are routed to people in the organisation who are responsible for addressing those issues - the pricing guys, the product, guys, the network. They get these elevations and they're responsible for acting on it. You hold them accountable and you measure them," Markey said.

Successful deployment of the third loop creates an integrated way of collecting feedback systematically and putting it to use, interacting with customers, and solving larger scale issue.

Tech led

None of this happens without a technology platform, says Markey.  "The ability for example, to trigger a request for feedback at a particular moment in a customer's experience, collect that, deliver it to the right person in the organisation at the right moment, then put it into an analytic platform so you can then identify patterns and issues - you need some technology to do that.”

Developments in technology have made it all possible he said, and platforms like Qualtrics are purpose-built for it.

“When we first started, natural language processing wasn't a thing, it wasn't widely available. The advent of text analytics allowed companies to address open-ended unstructured text at  scale, something that was previously very hard to do,  even if at small increments individually it was very effective."

But to address systemic issues you need to look over time and all these comments. That, in part, was the rationale for why in 2021 for Qualtrics paid US$1.25bn for Clarabridge.  Mark Bishof, the former CEO of Clarabridge, noted at the time, “Clarabridge’s ability to help companies discover what their customers are saying about them across unstructured sources and provide meaningful, actionable insights is a perfect complement to the Qualtrics platform."

According to Markey, “Natural language processing revolutionised the approach at scale across global organisations."

LLMs take centre stage

Large language models now offer another leap forward, according to Markey.

"Natural language was dictionary and tree-based. And you had to adapt it to your company if you didn't actually have it in the dictionary then it didn't exist. It is hard to use in some ways, even though it saved a tonne of time and energy, and it made some things possible at work," he said.

Now LLMs can be deployed to construct a dictionary. "Not only that, you can tease out patterns and themes from large quantities of vectorised text," Markey said.

This enables organisations to create their dictionary-based models that are much cheaper to run and also much faster. “You get the benefit of both.”

In turn, that makes things like pattern identification or issue identification much easier.

"And now we can actually have the machine not only tell us, ‘Here's the incidence of this and it is growing’, but actually interpret in a way that's meaningful to a human, rather than just [providing] a statistic," Markey added.