The majority of people are accessing their money digitally, and the use of smart devices—be it phone, tablet, laptop, or even web-connected appliances with purchase capabilities—is growing exponentially. And the volume of transactions happening per second feels countless, and perhaps what's even more daunting is the amount of security required to handle such a thing.
If you consider that every device in the world, be it a phone or a smart TV, is a potential access point for hackers, the need for reliable security suddenly gets put into perspective.
Fortunately, the financial services industry is already on top of this. Many of the world's biggest providers are leading the charge by combining big data with machine learning (ML). Not only does ML make your money safer, it delivers a better customer experience. Let's take a look at four specific ways the financial services sector is integrating big data into everyday operations.
Get the latest big data updates sent directly to your inbox—subscribe to the Big Data Blog now!
The digital age has transformed the way fraud works—not just from people unscrupulously trying to steal, but also the security teams attempting to protect customer money. Today’s economy is run via online transactions and transfers, which means that for fraudsters, gaining access (usually by stealing someone’s identity or credentials) is the goal. They attempt this in a number of ways, from skimmers on PIN pads to malware transmitted online to brute-force hacks of accounts. On a macro scale, that data can tell a lot about the different parties involved; patterns can create expected profiles and, more importantly, identify when potentially fraudulent activity occurs outside of those expectations. While the finance industry can’t protect everyone at every transaction, they can at as both a safety net and firewall against these types of bad actors thanks to big data.
To properly process this volume of data, various transaction datasets—with additional information such as interaction events and customer behavior—must be consolidated. That means storing data in an appropriate repository, such as a data lake, and applying ML to efficiently crunch the data while identifying patterns.
Regulatory compliance has been an issue for financial institutions since their inception. But in the digital world, regulations have rapidly changed. In addition to working within a digital landscape, regulations have quickly evolved to get a handle on new issues such as an increasing amount of cross-border transactions and the rise of cryptocurrencies.
Because of evolving regulatory rules, big data benefits financial services by offering large-scale processing of data sets as well as the ability to enact wholesale rule tweaks that quickly enable process updates for compliance. The collection of big data is the foundation for compliance, as it provides real-time proof of adherence to regulations (or identification of issues). This will never change the need for a compliance department to oversee and steer such things, but it will streamline and consolidate involved workflows, as well as minimize human error on records. A prime example of this comes from Caixa Bank, which saved 60,000 work hours overseeing Spain’s direct debits process.
Similar to fraud detection, regulatory compliance requires bringing together multiple sources. On top of that, compliance also utilizes advanced risk models, and these must be generated quickly without creating any impact on other projects.
Any organization’s operations can achieve valuable improvements with big data, and the financial services industry is no different. Consider the steps along any workflow; externally, banks and organizations are looking at customer retention and activity on loans, special offers, balance transfers, and other types of financial offerings. Internally, these same organizations are looking for any sort of process improvement, whether it’s in HR, IT, marketing, sales, or any other organization.
Big data provides insights that lead to innovation. Let’s take the example of maximizing customer engagement. Big data can look at a customer transactional data and account history to identify purchase patterns, geographic locations, and other potential engagement triggers. With ML, models can be built to identify the customer needs based on this data and extend appropriate offers that maximize potential for engagement. For example, if the ML model determines that a customer is doing a bit of remodeling work by shopping at hardware stores and related businesses, it could trigger an offer for a home equity line of credit.
To get the most accurate view of a customer, as many sources need to be used, including licensed third-party data regarding outside factors such as demographic and geographic data. Data scientists will also need to build and constantly refine customer models while also looking at big-picture economic factors such as interest rates.
As a subset to both fraud detection and compliance, financial services firms are facing increasing pressure from governments specifically regarding anti-money laundering laws (AML). Money laundering is a different issue from purely fraudulent transactions, and laws and regulations targeting this sort of thing have a much wider scope, including tax evasion, public fund corruption, and market manipulation. Other elements involve concealing these crimes and any money derived from these actions.
For AML compliance, data must be ingested from extremely diverse sources (sanctions lists, legal data, transactions, application logs). Also, ML models need to look at known money-laundering methods across timing and context in order to flag items for further investigation. Merely working within established rules (such as a transaction threshold) creates black-and-white thinking to an issue with a lot of gray-area manipulation by criminals. This is where ML can truly add value thanks to models that evolve over time as criminal schemes become more nuanced and sophisticated.
A wide range of sources is required for AML compliance, including taking on datasets that have many combinations of structured, unstructured, and multi-structured data. Models have to be built to meet the latest regulations, along with constant updating to maintain compliance. Other elements include using tools such as graph analytics to reveal hidden relationships.
This post featured an up-close look at big data in the financial services industry, but big data and ML can provide the same types of benefits for just about any industry. To learn more, take a look at Oracle’s Top 22 Use Cases for Big Data. Covering manufacturing, retail, healthcare, and more, this ebook provides insights into the power of big data across multiple industries.