Four Data Strategies to Scale Real-Time Operations

For financial institutions to thrive in today’s highly competitive digital environment, they need to understand their customers at the deepest level, hyper-personalize their digital experience, and ensure the integrity and compliance of every transaction.

It’s also important for incumbent financial organizations to take seriously the competitive threat posed by a host of digital financial startups. These young companies aren’t weighted down by cumbersome and expensive IT systems. They use the agility provided by the latest technologies to deliver fresh, new consumer experiences. The way Uber and Airbnb blew up markets that seemed impenetrable, startups are doing the same thing in the financial world. Innovative FinTech and DeFi (Decentralized Finance) companies are disrupting established financial service firms across payments, lending, borrowing, trading, saving, and more.

To succeed in today’s marketplace, financial organizations need to use an ever-increasing amount of data in real time. However, all this data—from signing up more customers and adding new functionality to personalize their experience to increased mobile usage and fraud prevention techniques—is putting massive and growing workloads on traditional data systems. As a result, many financial institutions struggle to cost-effectively scale, and increasingly are experiencing latency issues and frequent system glitches and costly outages. Customers today have little patience for technical issues, regardless of where the fault lies. The regulators have also noticed these problems.

For example, several U.S. brokerages have had embarrassing outages this year due to surges in trading volume. In India, bank outages have been so severe that the central bank decided to halt the operations of several of the country’s banks for several months until they could prove they had solved their problems. To their dismay, the central bank prohibited them from growing basic business services, such as adding new credit card customers, causing a loss in revenue and impacting customer loyalty and satisfaction.

Traditional data architectures are not designed to handle the real-time scale and complexity of data generated from today’s financial applications. They often encounter high operational costs, unpredictable performance, inconsistent data, and availability issues. On the other hand, modern data platforms have been designed from the ground up to address these issues. These platforms incorporate the latest processor, storage, and networking technologies, and feature geographically distributed scale-out architectures better suited to deal with rapidly growing data sets. At the same time, these newer platforms are aware of the need to co-exist with existing legacy systems to ensure a seamless customer experience.

As more financial institutions begin to modernize their data architectures to provide digital services, here are four best-practice data strategies they need to consider to cost-effectively scale their real-time operations. These suggestions will help financial operations avoid data outages, make accurate business decisions, provide personalized experiences, and fight fraud.

Use Multiple Data Centers to Provide High Availability, Uptime

Today’s consumers expect financial institutions to function without interruption. Any period of downtime can cost millions, if not billions, while also damaging their reputation.

To protect against downtime, financial institutions should design their data architecture to span multiple geographic reactions, whether it’s data centers or clouds, where each location is active. The data should be replicated across locations so that it can be processed at any location. If a local disaster takes down one data center or cloud, or if there is a hardware failure, operations automatically failover to another available location without losing data.

Ensure Data Is Consistent to Make Accurate Business Decisions

To make accurate business decisions in real time—such as detecting fraud, delivering personalized experiences, and offering fast, secure digital payments—financial institutions need to ensure the integrity of their data. The data that powers these applications and systems needs to be correct, complete, and available.

In the past, financial institutions often had to make a trade-off between data consistency and high performance. The traditional data platforms they used could provide strong data consistency to ensure that customers saw the same data across different devices or channels (think ATMs and mobile devices), but couldn’t necessarily deliver this real-time performance at scale. Modern data platforms are designed to keep data updated consistently across different systems and data centers at millisecond speeds, guaranteeing no stale or lost data.

Pull In Data from the Edge to Deliver Better Personalized Experiences

In today’s fierce market, customers are looking for that real-time, personal touch, which according to a study by Chase, most consumers require in any digital interaction. One way to deliver these types of experiences is to build 360-degree customer profiles. This can be done by implementing edge technology, where data is processed closer to its origin, which reduces latency and response times, as data doesn’t have to be sent across long routes to data centers or clouds.

Financial institutions can then aggregate data in real time from many different sources at the edge, such as mobile and location-based services, and then apply artificial intelligence (AI) and machine learning (ML) analysis to serve up the right service at the right time. With these real-time customer profiles, financial institutions can create personalized offers for pricing, credit scoring, interest rates, fee offers and loyalty programs, accelerate customer onboarding, and predict/mitigate customer churn.

Perform AI/ML-Powered Data Analysis to Fight Fraud

In addition to providing targeted recommendations, financial institutions can also use AI and ML systems at the edge to improve fraud detection by instantly verifying identities and stopping fraudulent transactions while not inconveniencing customers.

To surface fraud patterns successfully, AI/ML systems need to be trained and updated constantly to ensure the models are optimized to incorporate new data or customer behaviors. For example, fraud models developed pre-COVID-19 were trained using historical customer payment behavior, both in-store and online. As COVID forced many customers online, their buying and payment behaviors changed, rendering many existing fraud models obsolete.

Modern fraud systems have moved on from older, rule-based systems to AI/ML systems, with the most sophisticated companies now using neural network/deep learning algorithms. These systems need to ingest and process terabytes and sometimes upwards of petabytes of data in real time. The business and technical sides of an organization must understand the real-time performance needs for these systems so they are designed right the first time and are built to handle the newer AI/ML techniques that they may incorporate in the future. This will ensure that they have future-proofed their platform for use today and tomorrow.

Stuart Tarmy is the Global Director of Financial Services Industry Solutions for Aerospike. He has over 25 years of experience as a General Manager and head of sales, partnerships, and product management for leading global financial services technology, electronic payments, eCommerce, AI/ML, data privacy (GDPR, CCPA), and regulatory compliance companies.