Financial services need to use their unstructured data better

Using machine learning to analyse unstructured data offers numerous benefits for financial institutions, but why is ML adoption still so low?

Data is fast becoming the lifeblood of enterprise. In every industry, businesses are looking to leverage their data in new ways, and with the introduction of new technologies, businesses now have access to more data than ever before.

The financial services industry is no different. Currently experiencing a period of rapid change, disruptor banks such as Monzo and Revolut are causing the more established banks to re-evaluate their operations and processes so they can continue to provide unrivalled customer experiences.

Yet, while banks continue to adapt their processes, they have largely failed to utilise their data to its fullest potential. A recent Cloud Technology Solutions report, ‘The Future of Finance: How ML Will Disrupt Britain's Biggest Industry', suggests that financial services firms have huge amounts of unstructured data available that remains unused and unanalysed. And, only three percent of financial services companies are using machine learning (ML) to gain actionable insights from unstructured data. This represents a vast avenue of potential that financial firms have yet to tap into that could help them generate new advantages.

Benefitting banks

Sacha Giese, Head Geek at SolarWinds, points out that "tomorrow's banks cannot be run by yesterday's systems. Banks and their IT pros need to move with the times and future-proof their businesses with emerging technology, such as artificial intelligence and machine learning". If financial services can effectively implement ML, they could profoundly change the future of finance and what it can offer customers.

In particular, Daniel Cohen, Director of Fraud & Risk Intelligence at RSA Security, highlights how the use of ML to analyse customer data can help banks improve their fraud detection capabilities, "ML systems are capable of analysing huge amounts of historical transaction data to identify fraud patterns and predict if a new payment is likely to be fraudulent". Cohen continues, "By using the data they already have about a customer, they can reduce the need for constantly requiring new authentication for payments". By minimising the need for constant re-authentication, businesses would be able to offer a more seamless experience to their customers, helping foster deeper relationships and encouraging brand loyalty.

ML-augmented anti-fraud systems would also mark a true departure from previous approaches to fraud detection which have largely operated on a reactionary basis. Previously, anti-fraud systems have prevented future transactions after noticing potentially fraudulent activity. ML-based systems can look at data and transactions in real time and ask, ‘is this your normal pattern of behaviour?'. If transactions fail to satisfy this query the system can prevent it from going ahead, potentially saving an individual's finances before money leaves their account. And, thanks to the stronger analytical processes enabled by ML, the prevalence of false positives can be drastically reduced, helping prevent unnecessary disruption to customer experience. 

To continue reading this article register now