Jim Barnett of Aspect Software
Q Why is speech technology so important to contact center applications and what are the major benefits?
A Many customers' primary impression of a company comes from its contact center. Speech technologies can greatly improve this impression by offering a simpler, more powerful interface than existing DTMF-based systems. In addition to improving customer satisfaction, speech technologies can save companies substantial amounts of money by automating routine tasks.
Q What have today's customers come to expect from contact centers and how does speech technology help you meet some of those demands in the contact centers?
A Judging from my friends and family, customers expect impenetrable jungles of DTMF menus and long waits in queue from the average contact center. As a result, they dread calling up and resent the time that they waste. The judicious application of speech technologies can simplify the menus, lower the frustration level, and even speed things up by automating more of the process.
Q What are the "low hanging fruit" types of applications that should be "speechified"?
A Small company or department-level auto-attendant has always struck me as the most obvious application. In general, any application with a branching factor of 10 to 200 (that is, where the caller has between 10 and 200 choices at some point in the call) is a very good candidate for speech. Such applications are too complex for DTMF, but usually represent a fairly simple recognition task.
Q Describe a few surprises that you have witnessed when implementing a speech application for contact centers.
A The biggest surprise to me was how many applications, especially early ones, hadn't been designed from bottom up as speech applications. A couple were close to "If you would like to speak to sales, use your tongue to press 2". The hardest part of the deployment was convincing the customer that they needed to redesign their call flow.
Q Provide us with your thoughts on the various standards that are being implemented and discussed. These include SALT, VoiceXML, SIP, XHTML+Voice and others you see as important.
A VoiceXML is generating a tremendous amount of interest. It isn't clear if it will completely displace competing technologies (e.g., proprietary GUIs for workflow development), but I expect it to gain and retain significant mind- and marketshare. SIP has replaced H.323 as the basic VoIP standard. I expect it (and VoIP) to become pervasive. Customer calls will be coming in on the PSTN for the foreseeable future, but I expect contact centers to convert to VoIP internally. SALT is basically a desktop technology. I don't see it having a major impact in contact centers.
Q What are some limiting factors in the growth of speech technology and what should be done to offset these factors?
A Over the last few years, speech application development has moved from being an art to a craft (an art is something only a few geniuses can do - a craft is something many well trained people can do). However, there currently aren't enough skilled craftsmen to ensure that all applications get built well. But the number of craftsmen and their level of skill is increasing steadily, so this should cease to be a bottleneck.
Q Describe a successful speech technology implementation and why you thought it was successful. Please include any benchmark statistics that support your thoughts.
A One of our recent implementations uses speech recognition and natural language processing to allow customers of a major airline obtain information such as flight departure and arrival times. I view this application as a success because it has made it simple and straightforward for callers to get information that would be difficult, at best, to get using DTMF. I don't have statistics for this implementation, but similar applications have reported a 50% decline in opt-outs to the operator (compared to DTMF implementations).
Q What is your estimate (in percentage terms) for the number of contact centers in the US and Europe that are currently taking advantage of speech technology?
A I would estimate that half of our installed base is currently using speech and the other half is planning to implement it within a couple of years. We're not seeing any new deals that don't involve speech.
Q What type of CAGR do you expect in the next five years for speech technology applications for contact centers?
A The industry estimates that I've seen are for a CAGR of 70+% over the next five years. I'd expect the growth rate to be lower than that this year due to the lingering effects of the recession, but to more than make up the gap in 2003-2004. It is important to note that many contact centers have done trial deployments, speech-enabling only one or two of their applications. Therefore, a lot of the growth will come from such contact centers expanding the technology to all their applications.
Q What vertical markets have you seen the highest penetration of speech applications in their contact centers?
A Financial services and telcos were the first industries to adopt speech recognition on a large scale. More recently we have seen a significant increase in adoption rates in the government and healthcare sectors.