Artificial intelligence can help banks retain customers


20 October 2017


The use of artificial intelligence (AI) in financial services is growing all the time but is dismissed by some as purely a gimmick. However, AI offers many exciting possibilities in the banking sector, potentially boosting efficiency by reducing the need for human customer service operatives, as well as ensuring customers receive a consistent experience when they need support. Sophie Guibaud, vice-president for European expansion at Fidor Bank, explains how AI will boost banks’ efforts to attract and retain customers.


Artificial intelligence (AI) is a term that you will hear being used in a wide variety of contexts right now. From scaremongering tabloid articles about how the human race is doomed to be wiped out by robots, to genuinely useful tech innovations like Amazon’s Alexa, AI is everywhere.

Researchers at Stanford University define AI as “the science and engineering of making intelligent machines, especially intelligent computer programs”. The term can encompass chess-playing computers, Google’s search and autocomplete algorithms, and even automated systems that help people overturn parking fines. You’ll also often hear the term ‘machine learning’ being used to describe the same principle; strictly speaking, AI and machine learning are not quite the same thing, but the two are closely related. Stanford calls machine learning “the science of getting computers to act without being explicitly programmed” but, for the purposes of this article, we can assume the terms are interchangeable.

AI is beginning to make its way into the world of business, too. It’s becoming increasingly common for customer support services to involve AI systems that can be used to answer the most regular types of queries, whether this is on an automated telephone line or with online ‘chatbots’.

We’re even beginning to see examples of AI being used in the world of banking. A wide range of fintech companies and banks, such as RBS, Azimo or Revolut, have been quietly working on their AI assistants, which they have been rolling out on a trial basis since mid-2016. Other banks are experimenting too but few will talk publicly about it. Again, these trials are largely focused on customer service, but AI has plenty of potential beyond this realm. In the future, we’re likely to see banks using it to actually cross-sell and upsell products to customers. That calls for data to be collected and organised for the most basic chatbot functionalities, and then to be analysed in real time to provide an even better experience with a system able to learn and cross-sell intelligently.

The business case for using AI

There is a strong case for using AI within many businesses. While acquiring, training and paying salaries to staff comes at a cost, a system that continues to learn how to deal with customer enquiries with a decreasing need for human intervention will pay for itself in a short period of time. But it isn’t just going to save money – it can also ensure customers get a better support experience. While the number of human operators a business can have is finite (and they may not be available seven days a week, 24 hours a day) an AI-based system can handle as many enquiries as you can throw at it, no matter what time it is.

While, for more complex cases, a human operator will be required to pick up enquiries, AI systems will ultimately learn how to cope with most, if not all, cases. The more enquiries the system has to deal with, the better it will get at resolving them without assistance – that’s the theory, at least. This brings with it another advantage: a consistent experience for all customers. While people phoning customer support lines may already be unhappy, if their enquiry can be dealt with in an unemotional, efficient way that reflects a business’s best-practice guidelines, you can mitigate the risk of the customer taking their complaint to social media, where things could get out of hand. Human call-centre operatives are generally well trained to deal with difficult calls, but AI can ensure that every customer gets the VIP treatment, and no front-line staff get shouted at. IBM, which provides its AI platform Watson to RBS, predicts that the platform could even learn how to read a customer’s mood and adjust its answers accordingly.

The benefits of using AI don’t end there, either. With the detailed financial data that banks hold about their customers, it is also possible to use AI to create highly personalised profiles of each individual and predict their behaviour. While banks need to be very careful about ensuring that they have a customer’s permission to use their data in this way, it is possible to make relevant offers of financial products in a timely manner. For example, if the bank knows that a customer has just bought a car, it can instantly make an offer to them for a relevant insurance product. This enables the bank to increase revenues through either a direct sale or referral fee without a human salesperson needing to be involved – so the cost is lower, too. The customer also appreciates the extra value that has been added, and their loyalty to the bank is boosted.

But while banks have plenty of data about customers, many of them don’t exactly have it at their fingertips. Legacy IT back ends mean that data about one customer may be in several different places and not linked together in an intelligent way. Data management is traditionally a challenging area for banks, but the arrival of technologies such as AI should be a motivating factor when it comes to addressing this. Quite simply, the benefits of organising customer data such that it can be plugged into AI technology outweigh the costs.

Another reason that banks should be getting organised when it comes to customer data is the forthcoming arrival of the second iteration of the Payment Services Directive (PSD2). PSD2 demands that banks give access to customer data to any third party that requests it. This means rival banks – including the digitalfirst challenger banks – and other fintech companies could use this data to provide innovative services to the customer, displacing the bank that actually owns the customer relationship. In order for banks to mitigate the risk of other banks and fintech companies effectively ‘stealing’ these customers, they can get their house in order and use PSD2 regulations to their own advantage, boosting loyalty among their existing customers and winning new ones.

AI-based customer service

Businesses using AI for customer service typically use it for handling the most basic queries. These are questions that tend to come up time and time again, and have relatively simple answers. Though the questions may be framed in slightly different ways, the system is programmed to understand the vast majority of these variations and then learn the others. As time goes by, the system also learns to deal with increasingly complex enquiries, meaning the need for human intervention is reduced. Not only does this enable businesses to reduce their cost, it also results in a happier customer – response times are quicker, and the answers they get are consistently accurate.

You may be surprised to learn that ABN AMRO was using AI in customer service as long ago as 2001, with bots answering some of the most common customer questions. Today, the technology is much more advanced, with RBS’s Assist platform accessible to customers as an online chatbot. Such AI chatbots – also known as virtual assistants – can reduce waiting times, deal with common queries and direct customers to human operators who have the specialist skills necessary to answer their questions. In some cases, AI can act as an enabler for human operatives, arming them with data to solve issues more quickly, rather than having them resort to the frustrating trial-and-error-based scripts they traditionally use.

Cross-selling and upselling

While the use of AI in banking is currently in its infancy and limited to customer service functions, in the future, we can expect to see it being used in many new ways. The most significant will see AI being used in the role of a salesperson. Banks that do a good job of organising and interpreting customer data will be in a strong position to increase their revenues by identifying people who are open to being sold additional services. Whether it is the bank itself or an affiliate that makes the sale, there is a massive opportunity here that, on top of increasing revenues, will help banks provide better customer experience by suggesting proactively to the client what they need at the exact time that they need it.

While data has to be used responsibly and only with the customer’s express permission, there are many benefits for everyone involved to use AI in this way. Besides the example of the car buyer being offered an insurance product, consider a traveller who has bought tickets to fly abroad on holiday. They might receive a push notification on their smartphone when they arrive at the airport to notify them of the best foreign exchange deals so they can start spending money. Or a mortgage payer who has encountered an unexpected expense just before they are due to make their payment could be offered an emergency loan at a competitive rate. The bank has all of the customer’s details, and has already carried out credit and KYC checks, so the actual loan-application process would potentially take just minutes.

Different kinds of customers will have different needs that can be met through these AI innovations. Think of an SME needing to carefully manage cash flow in the run-up to Christmas. It may need to not only give its employees their pay packet early, as is traditional in December, but also buy extra stock in order to fulfil customer orders during this period. The bank could step in at this stage, knowing from past data just how much revenue the festive season will generate for the business, and offer a loan to finance the purchase of this stock.

The bank as a marketplace

This concierge-like service is likely to be how banking customers will encounter AI in the future. Because of PSD2 regulations, smart banks are going to be much more like a marketplace than they currently are, having numerous partnerships with other banks and fintech companies to guarantee the best possible experience for their customers. Banks are under an increasing amount of pressure to ensure that they remain front of mind for customers and don’t simply become a back end for competitors to take advantage of. The best way of doing this is to form multiple partnerships with fintech companies – as well as competitors – in order to offer a comprehensive set of services for all customers. These services can be presented to customers in a way similar to Apple’s app store on desktop or mobile devices – easy to navigate and showing many options to customers. They can get an emergency loan, download accountancy software, apply for a mortgage, check how much is in their savings account and make a payment to a friend from their current account all in one place, and in a matter of just a few clicks.

This experience will complement the AI technology as well. While the customer might not have the time to look through all of the available options and assess what’s on offer, the bank’s AI platform will be able to anticipate the customer’s needs and put the right services in front of them. The application and approval process will be quick and efficient; the bank will benefit from boosted customer loyalty and a referral fee; and the service provider will have found a new customer without having had to rely on an expensive marketing campaign or acquisition scheme.

The use of AI in banking is currently limited to customer service functions, but it has the potential to go further.