Skip to main content
A magazine by Pega

We'd prefer it if you saw us at our best. is not optimized for Internet Explorer. For the optimal experience, please use:

Close Deprecation Notice
AI gets personal
Paul Gary
Paul Gary
5 min read

AI gets personal

Marketers must learn how to think about the customer’s needs rather than their own. Verizon has been using AI to do that at scale, with impressive results.
Share this page Share via x Share via LinkedIn Copying...

“Psst. You want to buy a watch?” 

That, says Tommi Marsans, marketing technology strategist at Verizon Business Group, is how many companies today approach their relationship with customers. It’s an old-school strategy focused on pushing a product or service. It shows no regard for an individual customer’s needs and desires. “It doesn’t work,” says Marsans. 

She should know. Like many companies, Verizon used that approach for years. Even the personas and segmentation that the marketing teams used to try and fine-tune their messaging proved ineffective. 

As digital marketing promises more precision, these old-school approaches are embarrassingly inept. “They’re as far as you can go in an analog world,” she says. “The name of my particular segmentation might be soccer and chardonnay.” Characterizing someone as a soccer mom is obtuse at best.

AI to the rescue

Today, Verizon has a company-wide credo to focus on the customer and personalize every single interaction with them, from its marketing emails to its online self-serve portals. It’s able to do this with the help of artificial intelligence.

AI is still relatively new in marketing, but it’s gaining more traction. It excels at the kind of pattern recognition that people are good at, such as identifying pictures or speech. It does this using statistical models that it has built from analyzing mounds of historical examples. 

Marketers can use AI to find patterns in customer data too. It can analyze customer behavior and use these insights to make automated decisions. It not only improves the customer journey by delivering the sort of high-touch experiences they’ve come to expect, but boosts return on investment. 

Last year, 84% of digital marketing leaders told Gartner that that AI could help them deliver real-time, personalized experiences to customers. With 63% struggling to personalize their interactions, they could use the help, yet only 17% have turned to AI in practice. As the American Marketing Association noted in its report on the topic, “it’s no wonder the efforts stall out.”

...AI is best suited for learning things that require low skill, like pulling data from a customer's profile, their previous purchases, and past interactions across different channels. Unlike a person, it does this in real time, at scale...

Putting AI into practice

Marsans was determined to not let that happen to her team or to Verizon’s business customers. So, three years ago, with a basic understanding of how AI works, she began to build out the team’s capabilities. 

“The concepts are not challenging,” she says. Indeed, AI is best suited for learning things that require low skill, like pulling data from as a customer’s profile, their previous purchases, and past interactions across different channels. Unlike a person, it does this in real time, and at scale to make accurate predictions about the customer’s behavior and intentions. 

“That’s the part I’m really excited about; the predictive analytics,” says Marsans. “So, depending on four or five actions the customer has previously taken, the AI can pretty accurately predict what the sixth, seventh, and eighth actions will be. That’s tremendously helpful for us and them.”

AI in action

Here’s how it works. Let's say you’re a Verizon business customer and you have 15 phone lines but you’re no longer using five of them. So, you log into your self-serve portal to cancel those. 

As you log in, the AI, which knows you have five dormant lines, recommends how to repurpose them. You might choose to ignore those and go straight to what’s called the disconnect flow. The AI might then offer you a 20% discount to set up wireless business internet service on those lines. “The AI is creating more value and more possibilities for the customer,” says Marsans. 

Marsan admits that a decade ago, such suggestions from a machine may have seemed intrusive to many people. But today, your phone is your friend and the apps you use on it are your business assistants. People have learned to love the machine. 

That’s partly because AI offers better results. Verizon has a complex portfolio of over 3,000 products and services. “A rep is going to remember about seven of them,” says Marsans. AI can keep them all in mind to make more useful recommendations. 

Access to all the data also makes customer recommendations more consistent. Verizon’s previous messaging varied between channels. “I might get one offer in an email, I might see another offer in the portal, and I might get a third offer from the rep,” says Marsans. “There was no ecosystem connecting them.” 

Conversion rates improved, but she would like to see them go even higher by creating even more value for the customer. “You can't make the customer do something. The only way to do that is to create value through customer intimacy,” she says. 

AI can do that and more by creating seamless conversations that assist the customer journey. But there are lots of things it can’t do. Marsans says marketing leaders must keep this in mind as they look to adopt AI into their own marketing tech stack. Her experience implementing AI is telling. 

She recalls one AI model that the company trained to focus on high-margin recommendations. The offers that it recommended were so low - for example a $5 discount on a phone - that they made little sense.

Her team course-corrected to what it thought was a middle path but went too far, she recalls. “The AI didn't know what we were after, and we had it chasing volume instead of margin. And we were just giving away the house with these outrageous offers.” 

Marsans learned that they had introduced a certain bias into the AI. “It was this old product-centric bias that made it into our models,” she says. “We had to dial back on that a bit. What AI does well is what humans can do, but faster. What it doesn’t do well is make judgment calls.”

AI: Your customers' BFF

In the next few years, Marsans would like to see AI contact customers and help them find more value in the company’s products and services. Ideally, it would see gaps in what they’re using and make targeted offers. 

That might sound Orwellian, but she believes moving from a product-centric to a customer-centric approach will change the game for all enterprises in the future. 

“Focus on use cases that matter to the customer and that you can measure,” she concludes. “The customer is going to lead you in the right direction.” 

Share this page Share via x Share via LinkedIn Copying...