TD still years away from full AI integration, CEO says

Register now

When it acquired an artificial intelligence startup last January, TD Bank hoped to use machine learning technology to transform the institution, giving it the edge of a challenger bank.

But it's a heavy lift.

“What we didn't know a year ago when we brought them on was how different very large companies like us are [versus] these new nimble startups,” Gregory Braca, the $314 billion-asset bank's CEO, said in an interview. “We didn't know what we didn't know, but we quickly realized that the gasoline for the AI engine is data, and it's not just large amounts of data, it's a consistent feed of new data, usable data in an organized way.”

As a result, it's going to take longer than perhaps some expected to see the concrete changes the bank is seeking. Before that can happen, it has had to reorganize how it handles data.

"We are spending a lot of time, brain power and money on organizing our data to get to that end state,” Braca said. “We all want the next app and capability for our customers." But this effort "really is rewiring how the organization has worked.”

Braca said it will take time to fully see the benefits of the acquisition, particularly when it comes to using it for personalization.

"Personalization will have a big role. Customers want us to know them better, but this will be a journey that will play out over several years," he said.

TD Bank is not alone in facing a learning curve on harnessing the potential of big data and analytics.

Many banks are in a phase of digital maturity some call “transactional incompetence,” said Jacob Jegher, senior vice president of banking at Javelin Strategy & Research.

Digital transactions are easy, but banks have yet to use technology to regularly offer advice to customers or create a digital environment that helps customers avoid financial missteps.

Bankers themselves realize that the technology is mostly experimental.

“This is not, 'We got a 20% or 30% customer lift here or there,' " Braca said. “This is still really early days.”

But Braca fully expects the acquisition of Layer 6, along with an AI prediction engine and a team of 17 data scientists, will result in positive changes.

Where TD has executed on pilot programs, the bank has seen a lift in customer engagement, brand and loyalty metrics, said Patrick McLean, the bank’s chief marketing officer.

Last week, as the bank unveiled a new version of its Bank Human campaign, called Unexpectedly Human, its executives emphasized that it hopes to use its AI stack not only to increase loyalty among more established customers, but also to gain an advantage in reaching customers again after their initial interaction with the bank.

“We want to take that first instance of contact and make sure even that data is organized in a usable way,” Braca said.

Smart targets for fintech acquisitions will have multiple use cases, said Lane Martin, a partner in Capco’s banking practice. Martin often gets requests from various parts of the same bank to consider use cases for AI in that bank.

“The product group will work on developing that product to the best ability, but after that team completes its job, you want to see other applicability for that product,” Martin said. “Oftentimes large banks are stretched for resources.”

The AI startup that TD bought is a case in point.

While Layer 6 will take some time to integrate and bring personalization to TD’s customer experience, it is bringing other benefits to the institution.

“That startup is not just focused on the digital banking initiative,” Jegher said. “They are focused on advancing things in the health care space too. There can be multiple use cases to this, and not all of it is about integration.”

With so many factors to consider in AI applications, including regulatory and ethical concerns, banks can’t be faulted for falling behind with integration of the technology. Still, bankers generally are eager to start something new rather than having to spend time addressing back-end processes that have been around for a long time, Martin said.

Bankers also tend to tack AI onto a new project instead of asking where it can be applied across the bank, Martin said.

“It’s hard to eliminate a smart aspect of someone’s job,” Martin said. “But if you think about the applicability to the different functions the bank does, you can orchestrate a project and make more seriously material changes over time.”

For reprint and licensing requests for this article, click here.
Artificial intelligence Machine learning Predictive analytics Data science M&A TD Bank