The Big Data Takeover of Financial Services

The three vectors driving the need for big data methodologies in financial services are lucrative, hazardous, and mirrored (to a lesser extent) across verticals. In finance, the lure of prescriptive analytics, data science, and cognitive computing is fueled by:

  • Formal legislation: Myriad national laws compel operators in the financial space to combat financial crimes. Spurred by the tenet of Know Your Customer, financial organizations are forced to learn as much as they can about those they serve to prevent money laundering, fraud, and other financial misdeeds.
  • Regulatory adherence: Several regulatory accords dictate how financial organizations can use the data gathered for issuing product and services; some of these mandates apply to the use of data gathered to counteract financial crimes. Paradoxically, organizations are almost impelled by law to use big data techniques, yet exponentially incur risk for doing so.
  • Competitive advantage: The need to raise profit margins, reduce costs, and do so more effectively than competitors can is a horizontal concern that’s particularly acute in finance.

Although most organizations realize failure to sate these demands results in penalties and decreased market share, very few know fulfilling them can produce a profound profitability.

According to Cambridge Semantics VP of Financial Services Marty Loughlin: “There’s also a business opportunity that they’re missing by not truly understanding who their customers are, their relationships, and things that are outside the pure banking relationship itself that may impact a business opportunity with that customer.”

Truly knowing one’s customer—and managing alternative big data sources across business units for this purpose—catalyzes significant revenue generation across industries, not just in finance.

Alternative Data Sources

The rapid incorporation of alternative data sources alongside traditional ones is particularly cogent in finance because most organizations “pretty much have the same information today,” Loughlin noted. “They all have the same market trends, sales, and company performance data. They’re all running models off the same data; no one really has an edge.” Alternative data sources, however, provide that proverbial edge by equipping organizations with information others lack. According to Loughlin, traditional financial data sources include information organizations choose to make public such as production numbers and company reports, as well as market trends.

Alternative data sources, however, are based on big data and aren’t as easily attained or analyzed. Examples include satellite images of retailers’ parking lots, which might indicate micro or macro level trends throughout the industry. Weather data can presage effects on agriculture or factory production in regions of the world suddenly experiencing inclement weather. Social media and other means of sentiment analysis are also alternative sources contributing to what Loughlin termed alpha, “knowledge you have about potential performance the market doesn’t have.” Loughlin mentioned an airline use case in which on time arrival data was combined with social media information and stock prices so analysts “can take certain dimensions, do feature engineering, feed that to machine learning and see if…we can predict whether the stock is going to move based on their on-time arrivals.”

The Alpha Impact: Competitive Differentiation

The notion of alpha is gaining prominence throughout financial services because of its monetization capacity. Technically, alpha refers to “the profit gap,” Loughlin explained. “The market is predicting a certain level of performance based on all the information they have and if you can use better information to predict, say, higher performance, that gap is your specialized information. That’s alpha.” This concept’s also applicable within organizations, as illustrated by the transfer of wealth from parents to millennials. “It would be good to know that the person who opened a checking account with $250 in it could be someone who’s going to inherit a million dollars because of their parent’s wealth,” Loughlin remarked. “That customer is someone you may want to pay more attention to than someone who’s got $250 in their account and that’s all they’ll ever have.”

Organizations that truly know their customers and have the data management rigors in place to understand customer relationships can capitalize on internal data from traditional data sources. Couples may have a joint checking account, for example, but each spouse has individual products and services as well. Centralizing the big data management on the back end to understand how those products might relate to the joint account (and vice versa) can lead to offers for additional products and services for each person. Information about customer relationships may stem from internal or external sources, traditional or alternative ones. “It’s not just the customer 360 view from the databases and silos inside your organization, but it’s the ones outside as well,” Loughlin said.

Customer Lifetime Value

The millennial use case is important because it involves assessing customer lifetime value—a process that’s considerably improved by integrating alternative data sources with traditional ones in a centralized manner. In order for organizations to prioritize customers with a higher lifetime value than others, they need comprehensive customer overviews involving alternative big data sources. According to Loughlin, those sources “could be familial wealth, it could be where they work, it could be their educational background, it could be who their friends are, it could be social media. It could be a lot of pieces of information about that person.”

Predictive analytics and machine learning play an integral role in determining customer lifetime value. According to Jeff Lee, CMO of Seacoast Bank, which implemented a customer lifetime value model across its organization, that model features “a machine learning algorithm called an opportunity sizing engine. It’s combing through our entire customer base on a regular basis doing… analysis. So if I’m a CPA and I bank at Seacoast, and Rob is a CPA and he banks at Seacoast, it’s comparing the upper core value and the lower core value [to determine] what do we need to do to increase the value of that lower core value customer.”

Cross Departmental Opportunities

Implicit to knowing one’s customer, calculating customer lifetime value, and availing oneself of alpha is the ability to rapidly integrate, aggregate, and exchange relevant data across business units for the end user. Although this task is difficult for most firms, “Finance, like any other enterprise, has the trouble that their customers are in silos, especially in banks, where they’re in different business lines,” Franz CEO Jans Aasman observed. “Almost not a single bank can, with one query, see a comprehensive overview of everything that a customer knows.” There are numerous approaches for overcoming these silos, some of which involve cloud data stores, comprehensive data fabrics, data lakes, or knowledge graphs. Organizations deploying these mechanisms or others for horizontal department or business line access will almost surely identify opportunities for financial enrichment. Loughlin commented that it’s not uncommon for large financial institutions to have different customer databases for 401(k) plans and brokerage purposes. “Think about it, if you’re a 401(k) customer that’s heading into retirement and you’re going to start withdrawing money, [your 401(k) provider] would like to know that on the brokerage side, so that the money goes into your brokerage account, and not to someone else.”

The data management nuances of this use case are critical. Capitalizing on this opportunity not only requires linking together information between internal databases, but also doing so for timely action across business units. Lee mentioned that certain results from Seacoast’s customer lifetime value model are fed into its marketing department for automated offers. The ability to link data together in this example and in others noted above enables “more proactive things,” Franz VP of Global Sales and Marketing Craig Norvell said. “Perhaps providing new services to the customer or reaching out to them for a line of credit or things like that. Normally you’d be passive [about] that and need someone to come looking for it; instead, you can actually go out to them and suggest things.”

Identifying Alternative Data Sources

A key facet of obtaining alpha and exploiting it with contextualized understanding of internal and external sources is actually identifying the types of alternative data sources that add value for specific use cases. The legal rigors for Know Your Customer have propelled numerous financial institutions to subscribe to services like those provided by an extremely prominent media entity that’s “creating knowledge graphs around companies and the knowledge they can find around a company, whether that’s financial data, or in the news data, or whatever,” Aasman stated. “And the main customer of that is banks and financial institutions, because they need to know.” The intricacy of Know Your Customer becomes magnified for business or corporate customers, which oftentimes have several subsidiaries and various identifiers for different divisions. This reality is pushing the movement for Legal Entity Identifiers, which would provide universal identifiers for businesses to simplify this concern.

In other instances, the types of alternative data sources impacting an organization are dictated by a particular use case. Loughlin articulated a potential use case in which a nationwide, specialized sporting agency seeking to raise funds for a major international competition could identify company directors who played the sport in college, or who have children currently playing in college: “You’re starting to look at these connections, so you may reach out to Thomson Reuters about people who are on boards, you might do web scraping looking at people’s interests, or you might look at LinkedIn for where they went to college. There’s a lot of different information out there that you could bring together to build a profile of people who might be more likely to make a significant donation.”

Risk Calculations

Alternative data sources and the holistic understanding of one’s customer they fortify is a key component of minimizing risk for financial institutions. Similar to the way in which Seacoast is calculating customer lifetime value, organizations can also determine the potential risk of customers. The challenge in doing so, however, involves viewing a specific customer across departments and business lines—which has historically proved exacting in finance. “Personal banking, mortgage, credit cards, small business: those are all separate entities within various financial organizations,” Norvell revealed. Linking various databases across these business lines involves standardizing aspects of metadata management, business terminology, and data models.

Ideally, the more databases and sources (both alternative and otherwise) utilized, the more information organizations have to “start doing risk calculations,” Aasman posited. “In what situations are people defaulting on their mortgage; what are the signals.” Again, statistical Artificial Intelligence models are influential in the predictive and prescriptive analytics necessary to determine signals for risk. “If you look at the world as a set of events, you can say what are the signals that indicate if A and B happen, that C is likely to happen,” Aasman said. In finance, those signals are applicable to fraud, money laundering, or credit card payment failure. They’re also applicable to predicting churn, which is valuable to organizations across industries.

In Retrospect

Moreover, those signals are responsible for identifying opportunities for cross-selling and upselling, especially when utilized at scale with alternative big data sources for a complete understanding of one’s customers. The legal and regulatory demands of finance were the impetus for banks attempting to “have a comprehensive overview of everything that ever happened to their customer, independent of what business lines are serving him,” Aasman ventured. However, they also serve as the means of profiting from such intimate understanding of customers via the extraction of alpha, the determination of customer lifetime value, and the identification of horizontal business opportunities across departmental units. With parallels between these opportunities (and opportunities for risk mitigation) in financial services and other industries, these examples will surely be scrutinized and emulated in other verticals, too.    

Originally posted at AnalyticsWeek