Written by Sohaiyla Khalili
Imagine if everyone in your organization had real-time access to complete, up-to-date customer data. In banking, this capability translates to tremendous advantages. This is why globally, banks are embracing the potential of big data analytics to crunch all sorts of data—customer, regulatory, legal, risk, and other—to ensure compliance with regulations and get a better understanding of consumer needs and behaviors.
Differentiation through data
Analytics not only gives banks a better way to market their products—for example, by offering the right credit card to the right person or business at the right time—but it also allows them to verify information and streamline activities such as online applications, ideally making banking more efficient at a time when banks are striving to cut costs and compete with tech-savvy newcomers.
Analytics can also be used to develop unique services, such as using location information to identify bank ATMs in the vicinity, or voice identification to replace cumbersome passwords. The advances are promising—and necessary—but the hurdles can be daunting.
One large financial institution set out to give everyone in the organization answers to questions about customers in minutes, not days or weeks. Soon, the company found that the biggest stumbling block was the extreme difficulty of unifying multitudes of customer data sources.
Easier said than done
The bank had more than 100 large systems with petabytes of data across geographical locations and lines of business with two large external datasets. It suffered from low-quality, “dirty” existing datasets with inconsistent IDs and missing contexts.
Operating across multiple jurisdictions, the company is required to conduct risk assessment and compliance reporting for all its customer entities based on factors such as legal status, country of incorporation, and so on. Adding to the challenge, the bank has siloed data with fragmented legal, risk, and holding information, making the reporting process lengthy and costly with high risks of noncompliance.
The bank set out on a comprehensive project commonly known as know your customers (KYC). The main goal was to centralize the right data into one place where it could then be stitched together to produce a Golden Record for each customer. The trouble was that the bank had been using existing, traditional master data management (MDM) solutions that only use deterministic, rules-based approaches to data unification.
The company had attempted to use these solutions to stitch together all these systems in the past, but the complexity of the environment—the volume and variety of data—had proved too difficult to overcome. With an environment this large, writing and implementing rules often becomes too complex. Every incremental source that needs to be added requires hundreds of rules and copious amounts of time, making this approach at some point no longer feasible.
Adopting machine learning: a better alternative to MDM
To solve the problem, a team at the bank created a “data factory” approach involving a probabilistic, machine-learning based approach to data unification. In just over six months, the bank ingested and profiled 35 large data sources with 3.7 million rows of data to produce 325,000 clusters of customer records.
From a technical standpoint, through clustering, the company achieved about 40 percent record deduplication, increasing the accuracy of customer data substantially. Plus, the bank was able to attain more than a 90 percent matching rate with low Type I error to reduce non-compliance risks. The team can now onboard a new data source, from landing data to mastery, in about a week, and create a new Golden Record for every customer in a maximum of two days.
A better approach to knowing the customer
From a business standpoint, the bank created a unified customer database with deduplicated records and clear legal holding relationships. The new approach helped the company build a data pipeline that allows everyone to know each customer in minutes and match them across multiple languages.
The probabilistic, machine-learning based approach to data unification has yielded tremendous results in terms of time and cost savings. Even more important, customers benefit because the bank knows and understands their needs and can cater to them at every turn. The bank is now expanding this model into other important areas of the business.
This bank is just one example of a company that imagined the art of the possible and gained tremendous advantages by taking full advantage of its data. To learn more about the art of the possible and what it can mean to your organization, reach out to us or schedule a demo.
If you’re attending the Strata Data Conference in New York this September, we’d love the opportunity to demonstrate the art of the possible in person. Book a meeting with us here.