Josh Bottomley on big data & Ai initiatives
Expert Q&A: Banking on big data to enhance customer experiences
Last year, HSBC announced plans to hire 1,000 data scientists to work on AI and big data projects which it hopes will improve its customer experience and risk management. Josh Bottomley, Global Head of Digital Data and Development at HSBC, explains to LexisNexis® how big data is transforming the banking industry. He also warns that banks must earn customers’ trust over how their data is being used.
How and why is HSBC increasing its use of big data?
HSBC is increasing the use of big data right now around the customer experience. Clearly as a bank we have always used data—for example, we assess people’s credit and we are protecting against fraud. But the real focus now is giving people a customer experience that feels personal and relevant to them. By using a combination of data that we have historically about how people are logging in and what they are doing, we can now give an experience that feels more personal to them and keeps their money safe and secure.
How are banks more widely using big data?
In some cases, it is very straightforward—so for example we can see someone’s IP address and that helps us protect further against fraud. They are also using more sophisticated methods, whether it is advanced analytics or various forms of Artificial Intelligence (AI) and Machine Learning (ML), in other areas as well.
How does a bank benefit from using big data?
I think it will create a different type of banking experience. Banking used to be very episodic—you went to the branch or the contact center. By using data and then feeding that back to customers through messages or other personalized content, we can take the data into more bite-sized chunks, people can make financial decisions on the commute and start feeling more confident and competent about their money, and if we do that well I think it will be a really distinctive way of how we all engage with our banks.
What are the risks?
Clearly the biggest risk of data is around trust. As a bank we are obsessed about our customers’ security of their money and increasingly of their data as well. We now have this concept of data stewardship: we want to make sure we are really looking after customers’ data, that we are incredibly transparent about that and give customers as much control as they can around the data, but also that we use it for the right purposes.
We are very aware that customers sometimes make decisions without the information they need so some of the recent changes in the banking industry actually make it much easier, if you have got different bank accounts, to take the data and aggregate it, so we can help customers make more informed decisions across their whole portfolio. But trust is the critical issue—we need to make sure we build trust in our use of big data as opposed to undermining trust.
Are we seeing similar use cases across the financial services industry?
I think different businesses have slightly different use cases. I think the direction of travel in making customers’ experience more relevant and personal is probably a fairly broad one, but if a company is focused just on insurance versus one that’s in banking, they may concentrate on slightly different use cases.
At HSBC one of the things we want to do is use data that helps create a more holistic experience both for customers but also as we learn across different countries. We’ve got services that we’ve built in one country, potentially in an Asian market, and we will bring that to Europe, there are services in the Americas we want to bring. We want to be really quick to learn from those services and bring those across in the way that some of the really big consumer tech companies do already.
What sorts of datasets are useful to banks?
Banks have always used third-party data as well as internal data. We use credit bureaus for assessing credit, we rely on some third-party data to fight against financial crime, whether that is sanctions lists or PEP lists. I think we’ll always be using a combination of internal and external data.
But there are so many more datasets that we can use. For example, we’re about to test a service which will allow us to look at your behaviour of how you type and what you do as an additional fraud protection service. We are looking at other forms of third-party data—we were the first large UK bank with “connected money”, which is a way of aggregating your information across different accounts for the benefit of the customer. I think one of the advantages banks have is that they have been using and combining different data for many decades already and the technology is now allowing us to do that better and faster than we have before.
How does HSBC plan to use AI and ML?
We are already using AI and various forms of ML in different ways. One of the interesting ways we are finding is with chatbots on customer service—we have a service in the US, we have a chatbot in China doing customer service, and we are actually using it around some offers in pilot in Hong Kong. It is great because it is allowing us to understand questions that people will pose in very different ways using different languages, and to give them a very specific recommendation.
We use it in other areas as well - we use it when we’re worried about people who may be getting into debt in terms of communication. It is already a fairly established technology within the bank.
What is your prediction for trends around banks’ use of AI in 2019?
I think the big trend of AI will be cautious further adoption. The reason I say cautious is that actually the algorithms around AI, if you can get the data, are fairly well published and not that difficult to get hold of. The huge effort actually goes into getting the data ready, so the algorithms can work on them.
I think the other big trend we are going to see as a bank is around how we manage and control and assess the risks in communicating the use [of AI and data] to the customer to avoid that sense of “Big Brother”. At HSBC, we look at three big areas: we are very protective of customers’ money and their data, we also have a very clear view of data stewardship, and within that we have a very big piece of work around the ‘explainability’ of the model. So, there are cases already where we might use a slightly less good model from a predictive point of view because the results are more explainable to customers, because we want to be able to say why we’ve come to certain conclusions, and there are cases where we won’t use the AI because we don’t think we can explain it well enough.
The third area is helping customers with their own behaviors. I love people to say I bank with HSBC because my own financial behaviors are better, and we’ve used AI in support of that, where the customer feels in control of the settings or the level of involvement or engagement that they get from the bank.