Case studies

Business impact

Reduced the time needed for customers to make standard account inquiries by an estimated 30%, which allows the customer support staff to spend more time giving investment advice and less time giving account balances.

Issue

The financial services company wanted to improve customer service by allowing customers to quickly access account information without using lengthy menus or call center support.

Executive summary

By pioneering one of the first applications of IBM's Natural Language Understanding, the client increased self-service for the 401(k) application while making it more efficient and easier to use for their customers.

What IBM did

40l(k) plans were becoming more complex and difficult to manage with standard interactive phone menus. Many of the client's 1.3 million customers do much of their business over the phone. Already using menu-driven voice recognition, the company wanted to deploy Natural Language Understanding technology to improve self-service rates while maintaining high levels of customer satisfaction.

IBM's research and software divisions collaborated to develop a speech solution designed to understand normal conversational requests. The solution is based on the statistical Natural Language Understanding (NLU) technology pioneered and developed at IBM Research.

The IBM NLU approach is similar to the earlier pioneering work IBM did in speech recognition (now widely used in the industry). The data driven techniques IBM uses are more robust and capable than rule-based methods of handling noise and typical conversational style of speech, which may have hesitations or false starts.

The IBM solution was developed with the expert users or frequent callers in mind; however, for new or infrequent callers, it also provides a comprehensive context-sensitive self help feature to guide them. WebSphere Voice Response was used to answer the phones and WebSphere Voice Server supports the speech recognition and Natural Language Understanding. The end result is a system that very capable of understanding spoken customer requests and allowing them to manage their plans in a easier, faster way.

The solution was one of the first implementations of IBM's Natural Language Understanding when it was deployed to 60,000 customers in 2001. It was successful enough that it was extended to all of the client's 401(k) plans in 2002. Today, 1.3 million of the client's customers can get their account information as quickly as they can say "balance."

Capabilities applied

IBM Research has been a pioneer in speech technology for more than 25 years and has been awarded over 200 patents in that area. Combined with IBM's system and application development, Research's speech technology experience helped create a solution for the financial services company that brought new levels of service to their customers.