High-performance banking

How to improve marketing and tame risk with big data

Big data is an incredible opportunity for financial services companies. McKinsey Global Institute's June 2011 report, Big data: The next frontier for innovation, competition, and productivity, estimated that US banks and capital markets firms together had more than an exabyte of stored data in 2009. That much data makes a unified data management system – one with high-performance analytics – imperative.

Organizations must harness that big data to manage all needs efficiently. Inside those millions of transactions are the key to consumer preferences, red flags to fraud and money laundering, and the means to manage risk. The question becomes how to get to the data quickly and cost-effectively.

Today, many banks are using high-performance analytics to not only access the data but to make decisions that are increasing revenue, reducing risks, preventing fraud and improving marketing campaigns.

Mining transactions for marketing
Big data and high-performance analytics provide great opportunities to understand customers in ways that enhance and grow a business. Consider these examples:

• A large global payments firm uses predictive analytics to mine big data to support pricing strategy and decisions. The company analyzes 20,000 transactions per second across 17 dimensions to understand products, clients, products by client, and clients by product.
• One US bank uses data from 17 million customers and 19 million daily transactions as an early-warning system to detect customer disengagement. Certain interactions and transactions trigger alerts to front-line staff so they can contact the customer to nurture the relationship. After all, it costs much more to acquire a customer than maintain one.
• Another large US bank used analytics to shift from a product-centric approach to one that is customer-centric, and can now answer the question, "What is most relevant to my customer?" Analytics underpinning customer segmentation and campaigns is now funneled to support four actions: serve, offer, inform and retain. The goal is to understand what customers need from the bank, so each interaction will be positive. Predictive analytics mines customer and transaction data to provide insight into each customer's preference for products, campaigns, channel, contact and profitability. Analytics enables the bank to improve interactions with its base of more than 25 million customers.
• A European bank dramatically boosted both the number and quality of segmented campaigns – with big results. Combining its transaction data with demographic data, it was able to vastly improve its direct marketing. Sales from direct marketing grew from nearly zero to 30 percent in consumer loans, 33 percent in overdraft protection purchases, and 60 percent in credit cards. The profits totaled $30 million.

The return on these projects is real – and impressive. When banks use high-performance analytics to solve big data problems in marketing, they see big benefits.

Risk and regulation: managing with data
The risk side of banking is all about controlling costs and risk – including reputational harm, government fines, and market, counterparty credit or liquidity risks – that affect financial health or solvency. Data integration and quality remain paramount.

A risk analytics data model defines instruments, positions and counterparties along with market data, risk factors and models to compute risk exposures. It also supports stress testing and scenario analysis. With risk data housed in a unified repository, it is much easier to analyze market and credit risks, asset-liability management and liquidity risks. Aggregating risks across all portfolios provides a complete risk picture to the firm. This integration will help executive committees and boards of directors understand total firm exposure and how that compares to the firm's risk appetite.

One global bank implemented such techniques to perform regulatory and capital calculations and regulatory reporting at the group level. The bank processes more than 100 million rows of data per month, along with a reporting repository of more than 5 billion rows. The firm's single version of the truth now encompasses both risk and finance, helping to close the books faster. A unified data model and repository will help any firm meet the challenges of Basel III identified by the Global Association of Risk Professionals.

Full balance sheet risk analysis for assessing liquidity also demands integrating big data from multiple locations. Without integrated data, long calculation times can stretch to hundreds of hours, inhibiting timely decision making. An Asia Pacific-based bank tested high-performance analytical techniques to calculate a range of liquidity risk measures.

It analyzed a portfolio of 30 million complex cash flow instruments across 50,000 different scenarios in less than eight hours. The ability to fully revaluate liquidity risk nightly ensures informed funding decisions, even in times of market volatility.

What does that mean? By quickly determining exposure, portfolio value at risk, and liquidity coverage, the firm can determine products to take to market or markets to exit much faster. It can fine-tune responses to changes in interest rates, exchange rates and counterparty risk to remain competitive.

Imagine the advantage to a large US bank that reduces loan default calculation time from 96 hours to just four for a portfolio of more than 10 million mortgages. The bank can detect high-risk accounts much more quickly to forecast losses and hedge risk – plus make decisions about further lending.

Real-time risk assessment
The proactive element of understanding risk is critical. A large Canadian bank wanted to use 12 years of monthly account-level credit card data, credit bureau information and bank account information to better assess the risk before granting loans or raising credit limits. Ideally, it wanted this information in real time. To speed the computing, it used an in-database approach. When analytics works within the database, data doesn't need to be extracted, transformed and loaded. As a result, the bank could calculate risk 70 times faster.

With credit cards, proactive analysis can spot fraudsters before they run up thousands of dollars in fake charges – and stop them without inadvertently denying a legitimate purchase. A large global bank uses high-speed, real-time analytics to determine whether the purchase is legitimate at the point of sale.

The bank is so enthusiastic about the reduction in fraud losses that it has expanded the analytics to look at customers' online banking transactions. The intent is to build a more accurate profile a given customer's "normal" – and determine tip-offs that an account has been compromised.

The challenge for the years ahead will be to balance increased regulatory costs and the need for greater efficiency, while at the same time delivering an improved customer experience and innovations to retain customers and increase revenues. High-performance analytics will help bank executives pave the path to success – with customers, regulators and in the market.

Bio: David Wallace is the Global Financial Services Marketing Manager for SAS

David Wallace, SAS