The new face of banking

In the world of banking, a new breed of chatbots is making waves. Banking customers are today interacting more with chatbots than with bank executives or contact center agents who quickly add millions of dollars to a banks’ support costs.

Take Capital One’s EnoI , a chatbot that tracks bank balance, knows the details of customers’ recent transactions and even makes bill payments for them. You could ask Eno, “When is my credit card payment due?” using an interface similar to any popular chat applicationII and Eno will respond with accurate dates. You could tell Eno, “Make the payment now” and Eno will execute the order.

Scores of chatbots from other banks are already assisting customers – by answering questions about where the nearest ATM is located, to providing advice on how to improve savings.

Everyone’s excited about chatbots

Customers are excited about these chatbots. This is because they are already comfortable using technologies like Apple’s Siri, Microsoft’s Cortana, Amazon’s Echo and Google’s Assistant to complete everyday tasks. These chatbots -- powered by NLP, text recognition, machine learning, data analytics and other elements of artificial intelligence -- are accurately carrying out tasks such as re-setting Wi-Fi passwords and ordering groceries for users without human intervention. If chatbots can function in support and retail domains, they can do the same for banking.

Banks are excited about chatbots too. It isn’t difficult to see why. One reason is the rising cost of call center agents.

“Implementing chatbots…is a business imperative for organizations with extensive customer service needs. Chatbots will soon deliver customer satisfaction at significantly lower cost than human customer service agents,” says Gartner.

According to a study by a global consulting firm, a typical customer support call, using an agent, costs about US$4. Chatbots can reduce this to a few cents. Many banks can save as much as US$25 million a year by shifting 5% to 20% of call volumes to chatbots.

The other reason is that today’s bank customers would rather talk to an unbiased and fact-driven chatbot than to an agent who has inherent biases and obvious sales goals. Without losing on the feeling of connectedness that a conversation-based engagement creates, banks can optimize costs, increase customer satisfaction and improve outcomes. Non-intrusive, conversation-based engagements can be created in practically every banking function -- from client onboarding to money movement and customer service. 

Overcoming the challenges 

But banks have a few challenges to overcome before chatbots can become a reality. First, there is limited NLP support as banking domain-specific ontology is not exhaustive. The second problem is with security. Banks cannot have an open as well as secure integration with chatbots – and prompting a customer for authentication before a chatting session does not really provide a seamless user experience. Bank chatbots delivered over platforms apart from the bank’s own application will require adequate security measures with end-to-end encryption.

What could be the possible points of intervention to authenticate users better? Fingerprinting scans and biometrics will help create the required layer of authentication and security for in-house chat applications.

Spinning the right chatbot strategy

There are numerous approaches that a bank could take to its chatbots strategy (see Table1).

To begin with, there is the temptation to create a proprietary bank application. But there is no real reason to re-invent the wheel. The best option is to simply build the cart that sits atop the existing wheels: This means using third-party messenger platforms or using virtual assistants and devices to deliver chat functionality. Solutions on popular third-party platforms provide the best experience and acceptability of bank chatbots. When platformification of banking becomes a realityIV there will be a plethora of third-party apps which will use banks’ APIs to deliver financial services on these platforms.

This brings us to the question of the number of platforms on which a chatbot should be built. Should a bank build a chatbot for Google Assistant, WhatsApp, Facebook Messenger, Slack, Telegram, etc., so as not to leave out any of its customers? The answer is, “Probably not.” There are services like MyKai, Chatfuel, Beepbeep, Botkit and Smooch, which allow building and deploying a chatbot to multiple messenger platforms in one go. This helps maintain uniformity and context. It also makes it easy to synchronize content across platforms – a fact that is especially useful when a customer uses multiple messengers.

While chatbots, undoubtedly, have some way to go before we see mass adoption, there is no doubt that they will gather pace. For the immediate future, banks would do well to keep their chatbot strategy simple and invest in an AI platform that can power them in the future.

References

I. https://www.capitalone.com/applications/eno/

II. Like WhatsApp or FB Messenger

III. Predicts 2017: Artificial Intelligence, Whit Andrews, et al.; November, 23, 2016

IV. Driven by requirements such as Payment Services Directive2 in the EU

Suvendu Chand, With over 15 years of diverse global experience and an expertise in digital transformation, Suvendu has led multiple large-scale digital banking consulting initiatives in North America, Europe, APAC and the Middle East. He can be reached at suvendu.chand1@wipro.com

Khushboo Goenka Jain,Works primarily on conceptualizing innovative solutions. She has been with Wipro for more than 9 years and has worked on several digital banking initiatives in North America, Europe and APAC. She can be reached at khushboo.goenka@wipro.com.