The new weapon for banks to win the customer experience battle!



The bot as a powerful tool to improve customer service and operational efficiency


The competitive battle in the banking sector is now being played out on the customer relationship. The neo-banks are succeeding in winning over millions of customers, notably because they offer a fluid digital experience with a state-of-the-art mobile application that seduces users. This appetite for “all digital” was reinforced during the pandemic.

As a direct consequence of digitalization, customers are now looking for immediacy and efficiency in the processing of their requests. Bots (chatbots / callbots) best meet those expectations providing very high availability at costs far lower than those of agencies or call centers. A recent study, published by Juniper Research, indicates that staff savings from the use of chatbots in banking will reach 500,000 FTEs worldwide by 2023, or about 20% of the industry’s workforce.

But bots do not replace human account managers; they remove time-consuming or repetitive tasks from their shoulders allowing them to focus on higher value-added activities: advising clients on complex financial issues (savings, credit, insurance, etc.) or being attentive to their moments of truth (real estate project, professional or personal change, etc.).


Challenge #1: understand the customer’s intention!


The first objective of the bots, using artificial intelligence (AI), is to analyze the written or vocal messages formulated by the customer in natural language and to understand the underlying intention. The performance of a bot is primarily assessed by the percentage of correctly understood intentions (which will enable a coherent and relevant response to the customer). For a well-trained bot, this rate can reach 80%, provided that the Bot Mastering teams carry out a long optimization process based on Machine Learning technologies. As an example, the bots of the neo-banks Neon (N26) and Rita (Revolut) currently manage to process only about 30% of customer requests. The Bot Master exploits the contents of exchanges between man and machine to enrich the bot’s semantic field and improve the structure of dialogue trees. The multiplicity of formulations used by users to express the same need allows the bot to improve its understanding and gain in reliability.
That is why it is important that customers use the bot as much as possible. Therefore, the conversational agent (another name sometimes given to the bot) must be perceived by the customer as a “full-fledged” interlocutor. To overcome the reluctance that some customers might have to interact with a machine, banks will try to give it a personality and “humanize” it. Bots generally have an avatar and a name, speak in a simple and familiar language (no more banking jargon), are able to handle casual conversation not related to the object of the customer request (which is referred to as “small talk”)… Some bots are now able to respond with humor when challenged by customers about their status as a “machine” or point you to a good weather site when you tell them about the rain and the shine.

Everything must be done to ensure that the dialogue is fluid and that the user enjoys sharing a moment with the bot.


Challenge #2: provide the customer with an immediate and relevant response


Instantaneity is key. The bot must be able to respond quickly to requests; it must be trained to quickly recognize its limits and, if necessary, hand off a conversation to a human agent (via live chat or phone). Training a bot to process a hundred intentions with a satisfactory degree of reliability can take four to six months. It is therefore important to prioritize the intentions that you want to delegate to the bot (for example, the most frequent or the most impactful for the customer).. It is better to offer an optimal customer experience on these few types of requests, rather than aiming for exhaustiveness and not being up to the task. Who has never complained about a recalcitrant voice bot, which after X failed attempts still does not want to understand the key word you are trying to pronounce?
The quality of the response provided is, of course, just as decisive. Different resolution modes can be proposed by the bot in response to a customer request. The most basic mode is a “dynamic FAQ” type solution: it consists in providing the user with textual information, possibly completed by a video tutorial. The second level of resolution refers the customer to the application’s functionality that will allow him to process his request via a link or a clickable button. The top level consists in having the bot directly process the operation requested by the customer. For example, Sobot (Société Générale) is now able, in simple cases, to make a transfer directly in the dialogue interface.
On this point, the level of sophistication of the bot can vary greatly from one institution to another, depending on the maturity of the bot and the allocated budgets.


The bot, a key element in the banking relationship model


Much more than a gadget, a well-designed bot is a powerful customer relationship tool that allows banks to both stand out from the competition and improve customer satisfaction. Bots are available 24/7, are constantly in a good mood, eliminate customer waiting time, automate low value-added tasks and “educate” the customer by teaching him to use the bank’s products and services independently.
The bot is not a threat to human advisor: both are fully involved in a relationship model where digital and human are now totally inseparable and complementary. This concept of “augmented customer service” seems unavoidable for banks to improve the customer experience and remain competitive.
This is a significant but possibly progressive investment which is generally very quickly paid back.