In January 2017, a customer agent called Jim set a world record for paying a claim on behalf of the insurance start-up Lemonade. It took Jim just three seconds to cross reference the claim with the customer’s policy, run anti-fraud algorithms and send wiring instructions to the bank.
A superhuman effort. Literally. Because Jim is not human. He’s a bot.
Jim is one of an army of AI-powered conversational programs that’s changing the way people interact with financial services companies. Other companies who have launched bots include Mastercard, Wells Fargo, Capital One and many more.
Chatbots emerged out of a dissatisfaction with the way customers find answers to important questions. Historically, consumers had three basic options: visit a physical location; call an agent; browse a website/app. None is ideal. Conversational bots offer a compelling alternative. In theory, they promise the best of the human experience (natural conversation) and the best of digital too (instant and convenient).
Though ‘live chat’ agents have been around for a few years on desktop sites like Amazon and eBay, mobile has accelerated the concept. After all, the mobile is ‘always on’ and it is a far more natural messaging medium than the PC.
Facebook arguably kicked off the mobile AI era when it gave brands the ability to create bots for Facebook Messenger in 2015. Now there are tens of thousands on the platform. Other platforms followed. Bots are now supported across multiple channels such as Twitter and Slack. However, though these channels are popular, they are not universal. That’s why SMS has also become a critical platform for bots.
Indeed, bot messaging is now a growing piece of an enterprise SMS space worth $17 billion in 2016 and a projected $58.75 billion by 2020 (according to mobileSquared). The emergence of a next-gen form of SMS called RCS – which offers many rich features – should accelerate this process.
As of today, enterprises are highly interested in bots. Ovum’s Enterprise Messaging Survey 2017 found:
- 25% of organizations are already using bots to engage with customers.
- 92% of organizations using bots do so to automate customer-facing functions.
- 88% use them to reduce churn.
- 84% use them to reduce back-end costs.
- 72% believe bots are cheaper and more effective than apps.
- There could be 3.18 billion unique global mobile monthly active users of chat apps by 2020.
And, to repeat, financial services firms are leading the charge. One reason for this is that bank customers tend to ask the same broad group of questions:
- What’s my balance?
- How much did I spend?
- What’s this charge for?
These questions are, in theory, easier for a bot to answer than more open-ended queries (which are typically re-routed to a human agent).
While bots give customers speedy answers on platforms that are convenient for them, they also bring radical new efficiencies to banks. Bots don’t need a salary, holiday or insurance. They work 24 hours a day.
So, the savings could be eye-popping. “In the next 15 years, it’s likely that 45% and maybe up to 75% of jobs in the financial services sector will be performed by robots,” says Cliff Justice, KPMG advisory principal. “That should translate into enormous cost savings of up to 75% for firms that get on board.”
Juniper Research projects the savings for healthcare and banking could be $0.70 per interaction. That’s more than $8 billion per year by 2022.
But will customers resist? There’s no doubt that the stakes are higher for banks than others when it comes to AI. In a 2016 report, Forrester advised caution. “Money is an area where people are least willing to put up with mediocre bots,” it said.
“If Taco Bell’s TacoBot misunderstands a person’s request and orders three tacos instead of one, the results are unlikely to be catastrophic. If, however, a bot interaction accidentally leads to the same person sending money three times or paying a bill wrong, the results can be costly.”
However, early evidence suggests people are warming to bots. In 2016, Asia-Pacific bank DBS Bank created a bot for digibank, its mobile-only service. It recently reported that the bot now handles 82% of customer queries. It seems that, as new ‘challenger’ banks like digibank build customer experience around bots, caution will not be an option for the incumbents.
So, if consumers are ready to embrace bots then the question is: how much deeper can the relationship go? Can bots do more than just answer the same selection of common questions?
Many think they can. Indeed, when Bank of America unveiled its Erica bot in 2016, it said it was “designed to be not just a virtual assistant but each customer’s personal advocate.”
At the launch, BoA showed how Erica could not just show balances but detect spending patterns and then suggest different products that might help the customer avoid going into overdraft or make more return on their surplus. In this sense, Erica is not just a customer agent, but also a sales rep.
Meanwhile, Mastercard’s Facebook Messenger bot (available to issuer banks) lets users drill down into types of spending. Kiki Del Valle, Senior VP of Commerce for Every Device at Mastercard, says: “A customer can ask the bot how much she spent in restaurants in the last three months. She can then set a cap on spending and set an alert for when she nears the limit. It’s very easy to do this in a natural language chat session.”
Of course, the long-term success of bots depends on the naturalness of the communication. No one wants to be harangued by a know-all AI. And people will be unforgiving when a bot makes stupid verbal errors.
This may be why text-based bots, as opposed to voice bots, are a preferred option. Text sessions are asynchronous. Users can also store conversations and return to them when convenient.
Like after a huge and irresponsible shopping spree.