Here’s Why Martech Needs Some New Metrics

Once upon a time, a business sold a product — often a physical product — to a customer. That was a completed sale, concluding a marketing and sales effort.

Nowadays, thanks to technology — and especially the Net and related marketing technology — pre-sale is a journey, the sale is the beginning, and post-sale is the relationship. It’s not a handoff but a handshake, and one question is whether marketers’ metrics have kept up with this evolution.

“Fitbits aren’t just one-to-one,” a product for a person, personalization provider Amplero vice president Glenn Pingul pointed out to me.

They’re “me and the product,” he said.

Before you buy a car, marketing’s task these days is to put all the informational breadcrumbs in the right places so that you’re led to a sale. After you buy the car, it’s about keeping you as a happy customer, someone who shares the happiness with others — and buys again.

You need to “look at the customer [relationship] as a movie, rather than a snapshot,” Pingul said.

Suhail Doshi, CEO and co-founder of mobile/web analytics firm MixPanel, recalled that when his company began in 2008–2009, “the whole world was tracking page views.”

“We had to convince people that tracking page views was b.s.,” he said. Instead of such single-point measurements, MixPanel promoted the then-new idea of engagement, which tracks things like frequency of use and how often a user comes back.

That is, numeric values over time, appropriately fitting the idea that a movie is a rolling collection of snapshots.

“A Lot Of Different Moments”

Gilad Bechar, founder and CEO of mobile marketing agency Moburst, similarly noted that when his company began two years ago, “everybody talked about cost per download.” The more important measurements, he said, show the value of an app to a user over time.

“Your experience is a lot of different moments,” he pointed out.

So instead of point measurements like page views, email open rates, or click-through rates, many modern marketers have adopted various ways to measure engagement. Time on page or frequency of app use, for instance.

But, MixPanel’s Doshi pointed out, there are customers who might frequently use a product but not really like it. He mentioned Yahoo Mail. I’m thinking of a new printer I recently got, which I use because it’s not worth the effort to get a new one, but I’m not really happy with it.

Beyond single point measurements, or numeric samples over time like frequency of use, there are some qualitative assessments like Net Promoter Score (NPS). It focuses on whether a user would recommend a product, with the idea being that you wouldn’t recommend a product if you weren’t happy with it. Product recommendation can indicate product loyalty, a key goal of any marketer.

But, Amplero’s Pingul noted, we “still can’t figure out what behavior can give a [high score] on a NPS scale.” NPS is a measure of attitude at a point in time, but what happened in the movie before that point?

Loyal customers, the highest ranking, are both product owners and future product owners, which is why marketing now has such a wide span of involvement — pre-sale, during the sale, post-sale.

But, at any point, Bechar noted, “we have no idea if the user was happy with the experience.”

As it turns out, marketing technology may be pointing the way to the third phase of metrics, beyond point measurements and numerical measurements over time.

For instance, Apple Watch and other wearables can measure heartbeat, an indicator of excitement. A growing number of companies can automatically assess facial expressions for feelings. Sentiment analysis of social media is becoming more common, as is semantic analysis of hidden emotions in posts or emails. The burgeoning Internet of Things will mean that there’s no shortage of data points as you make your way through the world, some of which will be opt-in and identifiable.

How Do Happy People Act?

Perhaps “you can figure out a way to combine info about the customer,” Doshi said, “using machine learning [to determine that such-and-such] actions lead to a happy user.”

In the classic movie, “The Big Chill,” the girlfriend of a deceased character is asked whether her late boyfriend had been happy.

“I haven’t met that many happy people in my life,” she responds. “How do they act?”

That’s a hard one to answer. Not many humans who aren’t great poets can do so. But machine learning is designed to find recurring patterns in data that humans can’t otherwise discern.

“Instead of [trying to find] a metric that correlates with happiness,” Doshi said, “try to find with machine learning and neural nets the set of features that tells you a user is happy.”

Without machine learning to recognize the patterns, Pingul told me, “you can’t see the [whole] movie,” and you can’t predict what will work. His company employs machine learning to inform personalization.

As the Net makes product accessibility and price comparison universal, the quality of customer relationships — experiences over time — are becoming essential differentiators for brands.

Customer metrics can now become more qualitative, more continuous over time, so that marketers and brands can better understand the thing that any relationship partner needs to.

You know, like how best to stay together.