ListenLarger documents may require additional load time.
Thrifty speaks to its customers: car rental agency deploys speech recognition to improve customer experience while reducing costs
Customer Interface. 15.9 (September-October 2002): p30+. From General OneFile.
Full Text: 

Company: Thrifty Car Rental

Problem: Valuable agent time spent assisting "rate shoppers"

Solution: Installation of a self-service system incorporating speech recognition from SpeechWorks, based on a platform from InterVoice-Brite, a SpeechWorks partner

Deployment: February 2001


* More than 90 percent of customers say Thrifty's speech recognition system meets or exceeds expectations.

* Since February 2001, the system has handled up to 25 percent of Thrifty's total call volume.

* Agent efficiency is up 43 percent.

* Thrifty expects first year savings will total $1 million.

Thrifty Car Rental is the world's largest car rental franchise network. The rental firm prides itself on its high levels of customer service, but recently realized it had a problem: volume management. Its agent-based call center receives one million calls each year from "rate shoppers," people looking only to price rental cars. These inquiries were robbing Thrifty's call center agents of valuable time, tying them up with calls that could be handled more effectively through a self-service system.

After careful research, Thrifty chose to deploy a speech-enabled self-service system from SpeechWorks International to handle these calls, allowing rate shoppers to shop around while contact center agents handled reservations and other, more complicated requests.

The Scenario

About one million (40 percent) of the calls Thrifty receives each year are from people who need more information before they are ready to make a reservation. These callers want a quick, easy way to check Thrifty's prices before booking a vehicle. Thrifty, recognizing that having an agent answer each of these calls was costing the company both time and money, knew it needed to find a way to reduce costs in its call center. But Thrifty was not willing to sacrifice its high customer service standards to save a few dollars. An automated system would have to continue to serve Thrifty's customers above all else.

Many automated systems did not offer the level of service that Thrifty demanded for its customers. But SpeechWorks provided an attractive alternative: allow customers to shop for rates by themselves, having an agent step in when the caller is ready to book a reservation. The key was to ensure that the self-service system was efficient and easy to use for anyone who called.

Ease of Use for All Callers

To determine what system would work best for Thrifty, the company participated in the "SpeechWorks Here Guarantee" and the "Market Accelerator Program," two programs provided by SpeechWorks to help its customers achieve high levels of success with their speech services. The Guarantee ensures the delivery of effective speech applications, while the Market Accelerator Program provides customers with the necessary marketing services and tools to encourage high usage and ultimately, high return on investment. Based on the results, Thrifty elected to deploy a speech recognition system that runs on a platform from SpeechWorks partner InterVoice-Brite.

Unlike cumbersome touch-tone systems, Thrifty's speech-enabled reservation line, 1-800-THRIFTY, does not require callers to press an endless sequence of digits on the telephone keypad to access the information they want. Instead, callers speak naturally into the phone, answering questions posed by the speech-recognition system as if speaking to a live person. The system the answers provided and, after gathering all necessary information from the caller, can provide accurate rate quotes.

"Nearly 40 percent of our callers are checking rates and not yet ready to make a reservation," said Bob DuPont, staff vice president of reservations, Thrifty Car Rental. "SpeechWorks' speech recognition handles these inquiries quickly and accurately, which allows our agents to focus on revenue-generating calls."

Another benefit of the speech system over touch-tone is that the speech system offers barge-in capabilities. This technology allows callers to speak, or barge-in, before the system is finished speaking, saving the caller time and reducing the frustration of listening to long lists of choices before being able to select the one that is most appropriate. For example, an automated system may ask for the city where a caller would pick up the car. In order to provide the city name in a touch-tone system, the caller must listen until she hears the correct city to determine what numbers to press. With the speech system, the caller can simply respond "Chicago" to move on to the next step.

One concern Thrifty raised immediately was, "What if callers don't say exactly the right phrase when answering a question?" For instance, one caller might ask to pick up a car at Boston Logan Airport, while another might say Boston International Airport. SpeechWorks helped "tune" the system, allowing for such varying responses. The system, in essence, gets smarter over time, remembering the names or phrases that callers have used in the past to provide an accurate match going forward. With a vocabulary of over 70,000 words, including dates, times, car sizes, city names and airports, the speech-enabled system is able to understand almost anything a caller might say.

Another concern was that the system would be difficult to understand, speaking in a robotic voice. SpeechWorks' systems, however, use natural language technologies, providing a realistic, human-sounding voice. In fact, some callers have been so taken with the voice on the other end of the phone that talking to a real person is a bit of a let-down.

Thrifty's reservation line is also designed to take callers to the next step after checking rates. After the reservation information is gathered and the rental rate is quoted, the system asks the caller, "Would you like to go ahead and book that rate?" If the caller answers, "Yes," he is transferred instantly to a Thrifty sales agent. At the same time, the agent receives a screen "pop," displaying all of the information that a caller has previously provided. For instance, if a caller uses the speech recognition system to check rental rates in Boston in June and then decides to make a reservation, the agent's computer will bring up a screen telling the agent that the caller is looking to book a car in Boston in June. The caller does not need to restate the information already spoken. The agent simply confirms the booking and the customer is on his way.

Of course, Thrifty understands that some callers may prefer to shop rates the old-fashioned way. Those callers still have the option up front of dealing directly with a telephone sales agent. Further, at any time during the automated process, the customer can say, "Agent," and be transferred to a live telephone sales agent.

Customers Speak Up

Thrifty's speech-enabled self-service call center has been a hit with customers. A sampling of 1,000 callers soon after the deployment expressed positive experiences using the system, with over 90 percent saying that it met or exceeded expectations. And, since its deployment in February 2001, the system is up to handling 25 percent of Thrifty's total call volume. Callers have found the system easy to use and an efficient means to obtain the information they want.

Even better, agent efficiency is up 43 percent. What does this mean? Agents are converting 43 percent more calls into actual sales--a testament to the benefits of speech recognition technology. In fact, Thrifty expects first year savings will total $1 million.

"This is really a win for the customer and a win for us in that it provides a dramatic enhancement to the service we offer our customers while reducing our costs at the same time," says DuPont.

Source Citation   (MLA 8th Edition)
"Thrifty speaks to its customers: car rental agency deploys speech recognition to improve customer experience while reducing costs." Customer Interface, Sept.-Oct. 2002, p. 30+. General OneFile, Accessed 26 June 2019.

Gale Document Number: GALE|A91965840