Starting Small with Big Data Analytics

Fast-maturing Big Data tools provide companies with the ability to analyze far greater quantities and types of data in a shorter span of time. While retailers and technology companies have made significant developments on this front, the doors are starting to open for the financial services industry – paving the way for significant advances in areas from market research and customer segmentation to product testing, product development and customer service.

Perhaps not surprisingly, generating the right amount and type of data in the right format for analysis will be critical to economic success and demonstrating a return on investment. Too much data or the wrong data elements, as some organizations have already learned, may not always be beneficial. The key is to start small and take manageable steps toward incorporating Big Data analytics into your operating models.

For example, consider two of the most serious challenges facing financial institutions: retaining customers and satisfying expectations. Sentiment analysis and predictive analysis are viable options that can help banks effectively address these and other crucial issues.

Analyzing Customer Sentiment

Traditionally, consumer feedback has been collected using survey and focus group results. While these tools may gauge consumer sentiment, they may not necessarily capture emerging trends or hidden consumer desires, particularly on a real-time basis. Sentiment analysis aims to capture customer feedback from social media platforms and customer service interactions, among other sources, and provide insights with a shorter cycle between data gathering and action.

The idea is to use technology to create codes that analyze Web activity and provide insights into consumer sentiment on a larger scale and at a much faster rate than the findings revealed by traditional surveys or focus groups. Snippets of unstructured data can be interpreted and analyzed, delivering insights that can determine likely consumer response (both favorable and unfavorable) to the bank’s products, pricing, or service initiatives.

Consider the following customer statements:

• “ABC Bank’s small business offering is useful for new businesses and entrepreneurs. The lack of a same-day payment facility is a downer, though.”

• “The feature to view both business and personal accounts is really cool, although they really need to improve their customer service.”

The sentiment analysis tool would pick up words like “useful,” “lack” and “improve” and attach contextual meaning in order to generate graphs and reports, which can then be used by bankers to address customer expectations. Additionally, reports can be generated to illustrate trends and opinions on individual product and customer service features. This can help banks generate customer “wish-lists” and incorporate these into their product roadmaps.

Sentiment analysis can also help banks reward customers effectively. This is extremely important across the financial services industry as account switching costs are relatively low and customer churn is a major challenge. By examining customer confidence indices that are driven by specific data elements (such as product, functionality, content and price), banks can judge the mood of the market and decide how to best reward their customers. Successful execution in that drives loyalty and attracts new customers.

Although the technologies behind sentiment analysis are still maturing, many of the tools and techniques are advanced enough for financial institutions to derive incremental value by understanding customer likes, dislikes and preferences for product and service improvements. Early adopters are likely to gain a competitive advantage going forward.

A larger challenge is correctly predicting consumer needs and desires and responding, just in time, with an appropriate set of products and services. Predictive analytics mines large amounts of historical data to determine the likely occurrence of events in the future. By querying, visualizing and reporting these datasets, banks can illuminate behavioral and transactional patterns that can help with move-forward decisions on product and service strategies.

With predictive analytics, banks can build models based on customer spending, behavior and product usage to pinpoint products and services that customers might find more useful and that the institution can deliver more effectively. Such models can then be used to develop more efficient cross-sell offers, help banks increase their share of wallet, garner increased loyalty and lift profitability for attractive segments.

For example, profiling technology is helping credit card issuers identify transactions, cardholders and merchants that exhibit a high probability of fraud. By creating pre-defined profiles, banks sort through large data sets and identify fraud threats faster and with lower false positives. Furthermore, aberrant behavior patterns can quickly be identified, further helping to prevent fraud before it occurs.

As with sentiment analysis, additional research and testing is typically required to improve the accuracy and effectiveness of predictive analytic techniques. However, when deployed strategically, these tools can help banks gain a significant advantage in an environment with both traditional and new competitors.

Mr. Malhotra is a consulting partner and leader of the U.S. Retail & Commercial Banking Practice with Cognizant Business Consulting, a division of Teaneck, N.J.-based Co gnizant Technology Solutions. He can be reached at [email protected]. Mr. Jain is a senior manager and can be reached at s[email protected]. Mr. Kumar is a senior consultant and can be reached at [email protected].