September 1998

BY GENE EAGLE

Today’s large retailers face shrinking margins and growing competition, which means they must continually find new ways to cut expenses without cutting service to customers. Sears, Roebuck and Co., one of the oldest and largest retailers in the U.S., is meeting this challenge with an automatic speech recognition (ASR) system to handle calls that were formerly answered by live operators. The ASR system, developed with Nuance 6 from Nuance Communications and running on the Antares platform from Dialogic Corporation, has slashed Sears’ cost of handling each phone call to a tenth of the original cost. The system has also increased customer satisfaction by providing quicker, more efficient service than they were getting using the company’s previous system, which often required multiple transfers by human operators.

RINGING PHONES, SLOW SERVICE
Previously, customers would call a local number in one of the major metropolitan areas around the United States and be transferred to one of Sears’ central call centers to speak with a live agent. If the customer needed to speak to someone in a local store, the call center operator would transfer the caller back to that store. The system was cumbersome and inefficient, since a typical operator could talk to only one caller at a time. The phones rang constantly, and customers often had to hold for service. Equally troublesome to Sears, the ineffective system was expensive — costing approximately a dollar to process a single call.

After some investigation, Sears decided an ASR system would be able to handle the heavy volume of calls more efficiently and affordably than live operators. With the ASR system in place, callers could simply say what they are looking for — be it "tools," "women’s clothing," or "appliances" — and the system would quickly route them to the appropriate department in the store closest to the caller’s location.

HANDLING CALLS AROUND THE CLOCK
The Sears ASR system is based on the Nuance 6 speech recognition technology from Nuance Communications, which uses advanced linguistic and statistical models to interpret and understand natural human speech. The result is an ASR system users find as easy to speak to as a human operator.

The Nuance system runs on the Antares open DSP platform from Dialogic. The Antares board is a DSP expansion board based on the SCSA (Signal Computing System Architec-ture) standard, which provides features like distributed switching, logical addressing, and location-independent resource management.

Sears’ ASR systems comprise 22 speech recognition servers (a combination of Pentium Pro and Pentium II PCs running Windows NT) based in two call centers (in Ft. Wayne, Indiana and Golden, Colorado) that provide a total of 384 ports. Nuance equipped the Sears system with a grammar of more than 1,000 words, including synonyms, and the system’s in-grammar accuracy rate has risen to approximately 90 percent.

SATISFIED CUSTOMERS, LOWER COSTS
Sears began using the ASR system during the holiday season of 1997, every retailer’s busiest time of year, when the daily volume of calls can be double or even triple the normal level. Sears initially activated the ASR system for 130 of its stores, with 708 remaining stores using the old system. The ASR system worked so well that Sears quickly decided to implement ASR in all of its stores. Based on automated processing of approximately 150,000 calls per day, Sears has estimated the system paid for itself in three months.

Sears believes customer satisfaction has gone up with the ASR system because callers don’t have to spend time waiting to be connected to a live operator. The bottom line is that the ASR system met Sears’, and every retailer’s, goal: It concurrently saved the company money, and increased customer satisfaction.

Gene Eagle is a Speech Technologies product line manager for Dialogic Corporation. Dialogic is a leading manufacturer of high-performance, standards-based computer telephony (CT) components. Dialogic products are used in voice, fax, data, voice recognition, speech synthesis, and call center management CT applications. The company is headquartered in Parsippany, New Jersey, with regional headquarters in Tokyo, Japan, and Brussels, Belgium, and sales offices worldwide. For more information, visit the Dialogic Web site.