Even though most chatbots still only respond to simple questions, consumer use of such digital assistants is rising. This increased interest means it’s crucial banks and credit unions get in the game now. The intersection of voice technology and artificial intelligence is creating huge opportunities — and vulnerabilities — for financial institutions with chatbots. And the pace of change is accelerating.
Google’s Sofia Altuna, a key player on the team overseeing Google Assistant, says there are big shifts in consumer expectations when it comes to digital assistants.
“Almost overnight people have smart refrigerators, smart cars, and smart speakers,” Altuna observes. “This raises consumer expectations for your brand to be everywhere, and to be more personal, helpful and frictionless.”
Illustrating the potential for this AI-powered technology, Bank of America’s digital assistant, Erica, now has more than four million users, and in only became available to the bank’s full customer base in June 2018. The bank has 67 million customers, and 1 out of every 17 of them is using Erica. According to BofA, the top requests are:
- Searching past transactions
- Viewing account and routing numbers
- Sending money with Zelle
- Paying bills
- Managing credit and debit cards
- Contacting a specialist
The challenge for Bank of America and others going forward will be figuring out how to take such chatbots and digital assistants to the next level. Today they are primarily a means of responding to people’s queries. But in the future, they will need to be less reactive and become more proactive.
Instead of building something people come to — something that requires deliberate effort based on a conscious decision — financial institutions can create tools that take banking to people before they even know what they need, and can do so in the channels they want. Such highly personalized financial coaches could be so fine-tuned that some consumers may feel they are more aligned with their interests than another human.
This notion is very powerful, notes Ken Dodelin, VP of Conversational AI Products for Capital One.
Capital One was an early standout in this area, having been the first bank to launch a skill on Amazon’s Alexa platform in 2016 and, a year later, being the first U.S. bank to offer a natural-language SMS (text) intelligent assistant, called Eno. Customers can ask Eno questions via text just like they do using their voice with Alexa, and lets them make payments. But it also alerts them to when a payment is due, or that “You may have been double-charged.”
Learn how institutions around the country are saying goodbye to phantom growth with a new approach that creates real growth that lasts. Download eBook.
Explore a three-month view of consumer transactions and trends during the 2023 holiday spending season, including BNPL activity and mobile wallet purchase performance.
1. ‘Multi-Modal’ Will Be The New Requirement
The terminology of “conversational banking” can be confusing. “Chatbot,” “bot,” digital assistant,” “voice assistant” and “virtual financial assistant” are all related terms. Although some of the applications out there, including Alexa, Siri, Cortana, and Google Assistant are most often associated with voice interactions, they are increasingly “multi-modal.”
About half those who interact with Google Assistant actually use voice and some sort of touch-based input within the same “conversation,” explains Altuna. BofA’s Erica allows consumers to communicate by voice, text, or simply tapping buttons on the screen, with the latter being the most common mode.
Thinking “text-only” or “voice-only” or “desktop-only” is going to create problems for financial institutions, Dodelin warns. The fastest way for a consumer to communicate with a machine may be to just to blurt out requests like “Hey, Google, when’s my next mortgage payment?” But the fastest way for a machine to communicate with a person is not always to speak it, Dodelin notes.
“Listening to a bot list off your ten most recent transactions is painful,” he says. “You want to be able to scan them visually on your phone, computer or a smart device in your home.”
- Meet 11 of the Most Interesting Chatbots in Banking
- How Bankers Can Slice Through Chatbot Hype And Get it Right
2. Reactive Is Okay, But Insight is Better
While chatbots that can answer people’s common questions are helpful, experts agree that delivering people insights — telling them what the data actually means, for instance — is the future for such applications.
David Sosna, CEO of cognitive software provider Personetics, says financial institution will need chatbots that have the ability to identify and understand something that a customer didn’t know, then reach out to them in a message or a text, collecting more information along the way to ultimately guide them to a good decision.
“A standalone chatbot or digital assistant that doesn’t have the intelligence to reach out to consumers isn’t exciting.”
— David Sosna, Personetics
“Conversational banking is a new channel — another way to communicate with customers,” says Sosna. “A standalone chatbot or digital assistant that doesn’t have the intelligence to reach out to consumers isn’t exciting. But when intelligence is integrated into a chatbot, that’s a winning proposition.”
This is why Bank of America rolled out an enhancement for Erica this fall appropriately called “Insights” that will provide personalized and proactive guidance. For example, Erica can now send an alert if a consumer’s FICO score changes, as well as reminders of BofA credit card payments that are due. Down the road, Erica will send reminders for all bills — from both BofA and third parties.
3. AI + Voice/Text = Next-Gen PFM
Financial institutions have been pushing the personal financial management boulder uphill for years. Experts think artificial intelligence combined with conversational banking could be what finally fuels widespread adoption of PFM.
Peggy Mangot, SVP Innovation at Wells Fargo, feels the answer depends on how you view PFM. Typically it’s simply been a presentation of a person’s spending and budget, usually in a spreadsheet format that requires a lot of effort.
“The mindset is shifting to something more like a personal financial coach, or even a personal life coach.”
— Peggy Mangot, Wells Fargo
“But the mindset is shifting to something more like a personal financial coach — or even a personal life coach — because life and money are very entwined,” explains Mangot. This coach would be there in the moment helping a consumer determine, “Can I afford this new coat?”
A PLC should anticipate a consumer’s needs, says Mangot. For example, the digital assistant might say, “Peggy, you didn’t make as much last month, you’re going to need to start saving a little more now to cover your rent and utilities.”
This credit union redefined branch management. Manage one remote team and serve multiple locations. It's not just a solution, it's a revolution.
Software aside, your optimization strategy could be losing you money. But, with the right goals as your strategic foundation, your ROI will trend upward.
4. Make the Purpose Clear and People Will Trust a Machine
“Sometimes people are happier taking advice from a machine [software] than from a banking employee,” Sosna proclaims. “That’s because the machine does it without any other consideration besides the interest of the customer.”
The software can also offset consumers’ own worst interests and improve financial wellness — better than they would on their own. Sosna cites client bank examples where a consumer provides consent for the digital assistant to automatically move money into savings. The result was that savings balances with an AI-powered solution in place almost doubled.
“People are not very objective about their money and often will spend first before saving,” he explains. The software counteracts that.
Sosna also offers this useful anecdote: “When we tested the automated savings feature with a bank we found that the more customer goals and controls it offered, the lower the utilization by customers.” Eventually the bank removed the majority of the controls and goals and just kept the automation.
“We concluded that people are much more trusting of machines as long as they can clearly identify the goal of executing a very specific strategy, e.g. increase savings.”