How experimental tech drives TD Bank’s mobile app

Register now

While many banks have sought to employ experimental technologies when dealing with customers, including predictive virtual assistants, geolocation and advanced data analytics, few have brought all those pieces together to the degree used by TD Bank.

The bank has used such technologies live in production and won significant customer adoption, with its mobile app becoming No. 1 in the finance category in Canada for both iOS and Android. Mobile customers use the app 17 times a month on average, a figure that is growing.

TD Bank credits many of the fintech partners it works with, including Flybits for data normalization, predictive analytics and geolocation; Kasisto for conversational AI; and Moven for personal financial management.

“The vision is simple,” said Rizwan Khalfan, chief digital and payments officer at the bank. “We want to drive engagement and best-in-class experience. We started the journey five years ago. It’s been a combination of fintech partnerships and creating a platform internally. The partnerships are very deliberate, they’re addressing customer problems in a manner that helps customers live their lives, versus trying to promote a banking service or product.”
Following is a look at how the bank Is using such technologies and what other financial institutions can learn.

Taking care of data issues

Any attempt at personalization and predicting consumer behavior has to be grounded in accurate, well organized data. Otherwise the recommendations will be off.

Many banks stumble at this early step.

“When we talk to large banks, the digital executive tells us that two years or so ago, they started what they call a digital transformation strategy. They wanted to bring a lot of data together and engage with their customers on a more personal fashion, but after about six months, they realized that almost 90% of their budget is being spent on managing IT complexity,” said Hossein Rahnama, founder and CEO of Flybits.

When the executives came up with use cases, they couldn’t bring them to market because the data they required came from systems with different data structures.

"They would need to go build hierarchical IT teams just to be able to manage that infrastructure,” Rahnama said.

Flybits offers a drag and drop user interface for extracting data from different systems to provide intelligence about the customer, without having to integrate the data.

“We’re able to take our own data and contextual data that we can acquire on customers and connect those two sets of data without normalizing it into one metadata layer, and deliver experiences that are highly personalized and contextual," Khalfan said.

Using Flybits, TD draws customer data from its checking, mortgage, and insurance systems and combines it with contextual data — location and weather information, for instance — to offer useful advice to the customer.

Digital concierge

TD's digital concierge, TD for Me, has been used by 1.7 million customers in a variety of ways.

One is to help people interested in buying a home. TD has geofenced residential construction sites and homes for sale across the country. When a customer interested in buying a home is at one of those sites, they’re sent a preapproved mortgage offer based on the type of property they’re looking at, their credit score (pulled from a credit bureau), and their borrowing limits.

In a related use case, if a customer is walking by a branch and their mortgage is up for renewal within the next three months and a mortgage specialist is available in that branch, the customer will receive a message on their device inviting them to go into the branch to renew.

In these cases, the benefit to the bank is obvious: the customer might apply for a mortgage or keep the loan with the bank.

TD also uses this technology in ways that have no immediate benefit to the bank.

Partnering with Flybits and the public transit company Metrolinx, TD created a feature that provides train schedules and notifications to commuters. Intelligence is built in to sense, for example, what line the customer uses, so that they don’t have to select one.

“We removed a barrier for customers,” Khalfan said. “We can see you’re traveling the Richmond Hill line. Your train will be delayed by 15 minutes and it will be on platform 5. The first time customers got the notification they were delighted by it. They were like, 'Wow, that’s very useful,' because any of us that take public transportation, we time it down to the second. So that information allows us to plan our trip to and from home. That’s how we deliver an experience that’s relevant and contextualized for our customers.”

Twenty thousand TD customers currently use the feature.

“Some people say, why would you even do that?” given the lack of revenue or business case, Khalfan said. “Transportation to and from home is something we want to be a part of and make it easier.”

TD for Me also has an application for credit cardholders. If the app detects a customer is at an airport, the customer could be sent an offer for an upgrade, travel insurance or some other service related to their air miles card. The Flybits software can help predict the consumer segments most likely to respond.

TD for Me sends music fans notifications about upcoming festivals in which they might be interested.

PFM, chatbot

To help mobile banking users with spending and savings, TD partnered with Moven Enterprise two years ago.

So far, 1.8 million customers have signed up for the app feature, which is called MySpend.

TD has evolved the app in reaction to customer feedback, Khalfan said. One recent addition customers requested was the ability to see their spending in categories like dining, shopping, and transportation, and whether it’s above or below typical so they can make adjustments during the month to meet financial wellness goals.

On the virtual assistant side, TD piloted Kasisto’s “conversational AI” technology for several months and recently embedded the technology in its app.

Kasisto is a spinoff of the research lab SRI International, which also developed Apple's Siri. The first generation of Kai was a collaboration with BBVA. Thousands of calls at the bank were transcribed and fed into the artificial intelligence engine. Full-time writers at Kasisto produced additional dialogues for Kai and general banking knowledge, such as information about CDs, IRAs and credit scores, was also fed into the system.

To work with TD Bank’s customers, the engine had to be taught how to be bilingual, speaking in French and English.

Through the chatbot, TD app users will be able to check account information, review transaction histories and monitor spending levels. They’ll be able to get instant answers about spending-related questions, including how much they spent on recent weekend getaways, what their largest transactions were last week or what they spent on groceries or trips to the coffee shop last month.

TD Bank is launching the capability to customers in phases and by early next year the feature, called TD Clari, will be available to all TD customers.

Zor Gorelov, Kasisto's chief executive, noted that Clari is trained to teach customers what the chatbot can do and that TD plans to rapidly iterate on the virtual assistant based on customer feedback and conversations with Clari.

Clari will encourage customers to probe their financial life to better understand their finances and improve how they manage their money, Gorelov said.

"Instead of asking Clari, 'OK, tell me about my expenses last weekend,' TD’s customer could one day ask Clari conversationally: 'I want to go out to the theater this weekend. Can I actually afford it?' "

For all of these tech components, the sole objective is to drive engagement.

“What we’ve found is our engaged customers have the highest retention, the largest book of business with us, and the highest propensity to deepen the relationship with us," Khalfan said. "If you drive engagement, the rest of it sorts itself out.”

Editor at Large Penny Crosman welcomes feedback at

For reprint and licensing requests for this article, click here.
Mobile banking PFM Virtual assistants Geospatial data Predictive analytics TD Bank