May 10, 2018

Are banking chatbots hitting the trough of disillusionment?

disillusionment chatbot

This post was originally published in FinExtra 

There’s mounting evidence that the financial services sector is starting to come to grips with the chatbot hype. The recent decision of automated savings fintech Digit to scrap their chatbot made headlines, but was not entirely surprising.

What’s even more telling is that we are not seeing many successful widescale deployments of “conversational banking” in any form, and those that are available have rather limited capabilities at this point.

There are a few reasons chatbots have hit a rough patch in financial services:

Users don’t know what questions to ask

Chatbot engagement rates are low because people don’t know what questions to ask beyond the trivial “what’s my balance?” and “how much did I spend on food?”. They may want to check their balance before making a purchase, but beyond this – they don’t know what they don’t know and therefore are unable to effectively use the chatbot.

Conversation is a cumbersome way to deliver information

The idea that people prefer to ask questions rather than use a graphical interface has not stood the test of reality, and for good reason. Many types of information are easier to grasp when presented in a graphical format as part of the day-to-day flow of the user interaction with the bank’s mobile app or website. In the Digit app design overhaul, “the emphasis is on tapping on graphical interfaces to get information rather than forcing people to type in queries.”

Coming up with smart answers is still a major challenge

While AI has become pretty good at understanding user intent, most chatbots are only able to provide smart answers to a very limited set of questions. Accurately answering a question such as “can I afford to buy this phone?” requires the ability to perform real-time predictions that are far outside the skill sets of most chatbots and chatbot creators.

Beyond Chatbots: How to make AI work for banks and their customers

While chatbots can be helpful in certain interactions, many banks have quietly embraced other forms of AI that have proven more effective in engaging customers and helping them improve their financial lives.

Here are some examples.

Contextual interactions directed by the bank  

UK challenger bank Tandem has made worry-free banking its mantra. The Tandem Bank app will offer users personalized insights and advice to help them manage their money as part of their everyday banking experience.

According to Tandem, customers can expect to see a range of helpful insights –  “I’ve got your back” insights, such as unusual spending activity and tips to avoid fees; “Heads up” insights, such as a potential balance shortfall or upcoming uncovered bills; and “Get ahead” insights, such as opportunities to boost savings or investment.

The Tandem app aims to provide a conversational experience for the customer, but rather than using an external chatbot interface via text, it’s all happening within the app. It represents a shift in thinking away from conversations led by the customer, to contextual interactions directed by the financial institution.

Proactively engaging customer with smart conversations

Discount Bank recently launched its digital assistant Didi (which stands for Discount Digital). Like most banking digital assistants, Didi can respond to customers’ inquiries related to their account activity.

But it can do much more, proactively prompting customers with relevant insights and providing a forward-looking view into their finances. With these capabilities, the personal assistant is fulfilling multiple objectives:

  • Preempting service issues before the customer calls (call prevention instead of call deflection)
  • Providing customers with personalized insights and useful advice so they can better manage their finances
  • Increasing customer engagement and satisfaction with the bank:

While digital banking assistants are gaining popularity, the ones that will stand out will be those that deliver smart interactions based on true understanding of each customer’s financial behavior and proactively engage the customer with relevant and useful advice.

Automating money management

Last year, RBC launched Nomi Find & Save ™ and Nomi Insights ™, which provide personalized, timely and relevant insights to help clients manage their day-to-day finances through the RBC Mobile app. Both of these new capabilities use a client’s account activity to identify trends, unusual activity and potential savings opportunities.

As described by RBC, the process is entirely automated, so customers can save “without lifting a finger.” NOMI makes saving simple by using predictive technology to find pockets of money in a client’s cash and automatically moving that money into savings.

Making money management smart and easy works – customers have embraced automated savings with significant amounts of money already stashed in their accounts and a great deal of positive feedback to the bank.

The takeaways: Chatbots can be helpful for bank customers, if used right

It’s been easy to get swept away in the chatbot hype. It is just as easy to buy into the doom and gloom of “AI will never work.”

A practical approach to AI in banking is one that keeps an eye on the prize while watching out for potential pitfalls. Here are four things to remember as you try to sort out your AI strategy:

  • AI and chatbots are not synonyms. While a good chatbot will utilize AI, there is much more that AI can do beyond NLP and speech recognition. Successful solutions require an analytical brain that provides deep understanding of individual behavior and the ability to anticipate customer needs.
  • Chatbot is a communication tool. Conversation is a means, not an end. Conversational interfaces should be viewed as one of the tools in the bank’s arsenal for serving and engaging customers, and not necessarily the default solution for all interactions.
  • Focus your chatbot on specific customer journeys. Find the use cases where a chatbot can remove friction and can be quickly trained to be useful.
  • Consider other means of leveraging AI where chatbots fall short. Proactively engaging customers with AI-driven insight and advice as part of everyday banking can deliver a big payoff – customers improve their financial lives while simultaneously strengthening their relationships with the bank.

Big words and sexy demos are a far cry from real solutions that work and deliver value. While many chatbots are new to the scene, look for those that have a proven track record in widescale financial services deployments.