I’ve Got Algorithms
Executives at American Savings Bank will no doubt attest to those projections. Managers at the Honolulu-based bank decided early on to focus the company’s self-service efforts on an automated phone system. That’s because “when people want to do something quickly and easily, they tend to reach for a phone rather than a keyboard,” explains customer service manager Renee Lum.
In May 2004, the bank launched an interactive voice response (IVR) system developed by Dallas-based InterVoice Inc. The system allows customers to check balances, transfer funds, and handle a variety of other tasks by talking in a natural voice to an automated assistant. Unlike early IVR systems, which often sounded like robots with sinusitis (and which routinely missed or misinterpreted user responses), the bank’s system eerily approximates a human service rep. The digital customer rep — whose voice is supplied by a local Hawaiian voice talent — even greets callers with local salutations, such as “Aloha” and “Mahalo.”
Advanced speech algorithms enable the InterVoice system to recognize the phrases and sentences people use in everyday conversation. And like the highly popular “Julie” system at railroad operator Amtrak (see “Ernestine, Meet Julie,” Techwatch, January), a “barge-in” capability allows callers to interrupt the system, meaning experienced users can quickly get the information they desire. The system’s speech-recognition engine, which understands more than 100,000 words, can also separate a significant utterance from coughs, sneezes, dog barks, and other random background noises.
After the development work at American Savings was completed, the bank and InterVoice tested the application on customers of various ages, ethnic backgrounds, and occupations, soliciting opinions on its usability. “We wanted our customers to like the system and to turn to a human only when they couldn’t get something from a machine,” says Lum. She notes that within two months of its launch, the system was handling more than 340,000 calls a month.
Lum says the friendly, Hawaiian-accented voice of the digital service rep immediately places customers at ease. It also helps them forget they’re talking to a computer. “It has encouraged them to try the service,” she notes. Managers at the bank have received very few customer comments on the system. “That’s kind of a good thing,” Lum insists. “Any time they don’t like something, they’re going to scream and yell.”
Lots of ‘Bots
CRM experts say human service reps remain unsurpassed in assisting customers. Of course, they also cost more. The reality is, many customer-support functions supplied by a person can be replaced — at a fraction of the cost — by a virtual automated service agent (for a comparison of costs of various customer service approaches, see chart, this page). Such automated service agents, known as ASAs, are software ‘bots that communicate with customers through a pop-up IM program on a Website.
Harrisdirect, a Jersey City, New Jersey–based online discount brokerage, began using an ASA earlier this year. The system has virtually eliminated the need for rep-to-client hand-holding, while smoothly guiding users through complex investment, regulatory, and taxation issues. The technology for the system was provided by Conversagent Inc., a software developer located in New York and Sunnyvale, California. The company’s flagship program, Conversagent Automated Service Agent, is designed to handle routine inquiries, freeing a client’s support staff to tackle more difficult cases.