By Yvon Moysan1
1Academic Director, IÉSEG School of Management
According to Andrew Leonard, a bot is “an autonomous computer program supposed to be smart, endowed with a personality and which most of the time, gives a service”. The term “personality” is justified by the level of anthropomorphism of the bot. The assumed intelligence of the bot brings us to the notion of Artificial Intelligence (AI) proposed by Marvin Minsky and which consists in “creating machines that can make human tasks which require intelligence”. One of the well-known examples of this kind of AI is Siri, the numerical vocal assistant of iPhone developed by Apple. When this bot interacts on platforms such as Facebook Messenger or WhatsApp, we then talk about “conversational agent” or “chatbot” (a combination of “chat” and “robot”).
Chatbots aim to improve interactions with customers. Gartner already predicts that, across all sectors, 25% of mobile interactions will be conducted via conversational agents by 2020.1 Moreover, Gartner predicts that by 2020, individual customers will have more conversations with a bot than with his or her significant other.
Considering the advantages of chatbots and the positive predictions of development, combined with the current growth of mobile utilization, firms are tending to use more and more chatbots as part of their suite of customer relationship management tools and channels. Whether it is assistance for a purchase, managing a booking or answering a question, chatbots are used day and night with the aim of delivering an automated and personalized experience.
More and more banks are exploring the opportunity to let virtual assistants manage some of their customer support services, by providing extended availability at lower costs. For the banks, it is an opportunity to reduce the overhead related to call centres and, if needed, to redirect the customers to human advisors, when the questions are too complex or if they require a deeper level of expertise.
According to Juniper Research, chatbots will generate savings in excess of US$8 billion (EUR 7.3 billion) each year by 2022, against US$20 million in 2017, with the health and bank sectors being the first to benefit from these new customer service possibilities.
Juniper Research estimates that the average time per question will be reduced by over 4 minutes compared to a comparable call centre interaction, which translates to a saving of between 50 and 70 cents per interaction. This maximal average level should be reached by 2022 for the banking bots based on messaging. Juniper Research expects the success rate of bots’ interactions (without the help of a human operator) will exceed 90% in 2022 in the bank sector, and more than 75% in the health sector.
The long-term goal for banks in terms of positioning will most likely be the progressive increase of the use of voice assistants such as Amazon Alexa, at the expense of imbedded IM (Instant Messaging) in existing desktop and mobile digital experiences, which will face a parallel decline in usage.
The advantage for banks in these intelligent systems which can learn from their mistakes is evident: they enable them to answer customer requests via a straightforward communication channel, 7 days a week, 24 hours a day, and they can also expect a reduction of customer dissatisfaction by lowering operational costs.
French banking giant Société Générale offers customers who visit their Facebook Fan Page “par amour du rugby” (for the love of rugby) the possibility to interact with their chatbot. All Facebook Messenger users can now instantly access the daily match schedule program for the French Rugby Championship and the rugby match results as well as other exclusive contents. The goal of the bank is to test the potential of this new technology, especially this kind of relationship channel considered as easier, more personalized and instant.
In the same spirit, Société Générale established a partnership with JAM, a startup specialized in student guidance, which has designed an automated instant messaging application like “Artificial Intelligence for the good things in life”, based on tips to simplify the daily life of young people.
The goal of this partnership for the bank is to better understand the needs and expectations of people under 30. Being on first name terms, the chatbot will have the goal of redirecting young prospects to Société Générale for the bank to answer any questions these young people may have about money, with a target baseline for the accuracy and relevance provided by the algorithm.
At this point, some banks have entrusted part of their customer service to conversational robots to answer customer questions. Among them, the Royal Bank of Scotland (RBS) announced, in September 2016, the launch of a system with IBM to provide 10% of its customers with the services of a chatbot.
The purpose of this system is to ease the workload of the bank’s call centres by taking care of routine questions regarding matters such as address changes or credit card activations, and to redirect the customers towards advisors to take care of the more complex questions.
Spanish bank BBVA also offers its first “chatbot” (reachable by Facebook Messenger and in Telegram), which essentially allows customers to check the balances of their accounts and credit cards. BBVA also offers a second chatbot, available in all messaging tools using a personalized keyboard, which allows customers to send money to a contact in a very frictionless manner.
Considering that verifying account balances currently represents more than 90% of access to banking applications, and payments among friends are frequently used, banks of all sizes and profiles clearly need to be able to respond to these common use cases in an efficient and cost-effective way.
Bankin’, the FinTech company specialized in banking data aggregation, launched a chatbot running on Facebook Messenger in September 2018. This particular chatbot enables customers to obtain information on their account balance and their last transactions, as well as acting as a financial coach by suggesting, for example, that the customer save some of their available cash balance.
Overall, even though initiatives have proliferated in recent months, banks are still quite cautious, mainly because costs and time associated with implementing chatbot infrastructure are significant. In addition, for some customer interactions, chatbots are still unable to provide 100% accuracy, with the most frequent errors being linked to off-topic responses or simply being unable to answer.
It is important to be aware of the customer frustration such interactions may cause, and the potential consequences in terms of image for the bank.
Through these examples, we can observe two approaches of chatbots in the banking sector: “showcase” chatbots, whose main purpose is to create visibility around a brand, and “customer service” chatbots. There is still a long way to go but, based on progress so far, it is reasonable to expect an increasingly pervasive use of chatbots in the banking customer service arsenal of tomorrow.