© $YYYY SourceMedia. All rights reserved.

How TD Bank wants to use data to read its customers’ moods

While many financial institutions want to discover as much about their customers as they can, TD Bank is looking to go a step beyond most by assessing consumers' moods when they are interacting with the bank.

So-called sentiment analysis software is already used in call centers to help detect when customers are getting frustrated or angry. But TD Bank aims to deploy it to capture mood as people interact with internet-of-things devices, with virtual assistants, and with the bank's mobile app and online banking.

It's challenging, because mood detection has to be of the moment, rather than relying on analysis of what a customer did yesterday or last month.

“The journey for us is more about getting a holistic view of the human being,” said John Thomas, executive vice president and global innovation head for TD Bank Group.

The bank employs a combination of tech to make this happen, including chatbots, voice analyzers and phone channels.

“Some companies that are purely remote in their interactions will probably go down one technology route,” Thomas said. “Companies like TD that are both remote and face-to-face or phone-to-face will have more robust means of gauging mood over time.”

The effort is part of TD Bank's recognition that traditional digital marketing doesn't cut it anymore. Customers want personalized digital interactions.

“Traditional demographic based segmentation doesn't really reflect how people see themselves today,” said Thomas. “Our thought process, and it's validated by data, is these concepts of experience and perception with a brand are probably a better pursuit than pure conversion.”

Thomas has a sense of urgency about that mission. “We are seeing data emerge across multiple service verticals where people are now leaving service providers, citing a lack of personalization as a primary reason,” he said.

He’s not alone. In a study Celent released in January, 22% of bank executives said their companies provide personalized offers in their mobile apps and 57% said they are planning to do so.

“What TD and other banks are after is providing personalized and relevant insights at scale, delivered digitally,” said Bob Meara, senior analyst at Celent.

Mood analysis

Thomas noted that the human brain has a built-in mood detector.

“If you're dealing with a co-worker, spouse or family member, what you're doing in that moment is trying to understand the other person's context and their mood. And then you're essentially censoring yourself based on that mood.”

If he’s planning to give feedback or make a suggestion to his wife, for instance, it’s not enough that he knows her personality.

“I also want to understand where she is in this particular moment,” Thomas explained. “Because based on her personality, it might seem fine to propose something, give her a piece of feedback or ask her a question. But her mood may say ‘Not now’ or it may say, ‘This is the right moment.’ ”

TD Bank is trying to emulate that natural assessment in its interactions with customers. To enable it to understand moods and personalize interactions across all channels, TD is redesigning its data layer.

“We have seen many people in banking and other traditional service verticals try to use a big data lake to do this kind of interaction personalization,” Thomas said. “And that's a monumental struggle and a large investment.”

Data lakes don't work well for real-time interaction layers, he said. For one thing, data streams inevitably arrive in different time frames.

“We still live in a world where not all data is real time — it doesn't matter what service vertical you're in,” Thomas noted.

So the bank built what it calls a “fast data architecture” that sits alongside its big data architecture. The faster system will handle interaction across channels and marketing interactions.

"That might include all of our system of record data, like our product systems and platforms," Thomas said. "It might include our channel data and anything else that's relevant, but it's going to be made to inform when somebody walks into one of our stores or calls us on the channel as opposed to data architectures that are made to run big reports or business insight queries.”

Customer insights

The bank is developing insights about customers that can be provided to branch staff, through mobile and online banking, through direct marketing, and through phone banking.

“The journey for us is more about getting a holistic view of the human being,” Thomas said. “We serve that through this data layer and then use that same data layer to inform interactions, no matter how the customer comes to us or we come to them.”

This could include a view of the different transactions a customer has been doing in different places — in a branch, at an ATM, in their mobile app.

Every night, TD Bank analyzes its customer base and flags anything significant that happened with a customer during the day. The next day, it makes those flags and insights available to its interaction channels.

Some of the flags are warnings, as in, this customer might be leaving to go to a rival bank. They might be opportunities, like finding out a customer might be ready to buy a home.

“Sometimes something is happening in a customer's life and we want to be on the spot and relevant in the next interaction,” Thomas said. “Sometimes it’s a signal of a deterioration of the relationship or a warning sign. We want to be proactive about that. All of this is aimed at feeling relevant to you in that moment of time and being contextual as opposed to sell, sell, sell.”

Because of this work, TD Bank has been able to migrate from its traditional method of market segmentation around, say, propensity to buy something, to “relevancy triggers” where even if the segmentation suggests a customer might be interested in getting a credit card, an event in the customer’s life would tell the bank something else in their life matters more to them right now. This can inform marketing campaigns and in-person interactions with customers.

One example of a relevancy trigger: Among TD Bank’s mobile banking customers, it found a segment of people who frequently use the mobile app but go to a bank branch every other Friday to deposit a check.

“What if we took those insights and sent you a text message that includes a short video that shows you how to use mobile check deposit?” Thomas said. “That’s a little thing that’s not segmented but is just about the customer’s behavior.”

Algorithms help the bank connect the dots and find such opportunities today.

“Pieces of this journey will be AI-enabled over time, but there’s so much companies can do today to get to a segment of one without AI,” Thomas said.

For reprint and licensing requests for this article, click here.