Building the Case for Speech
Building the Case for Speech
By Donna Fluss
Speech recognition software really is the Rodney Dangerfield of the self-service market place. It gets no respect no matter how good it is. And it is good today.
Speech recognition is compelling technology because it appeals to the masses it interacts in the most common form of communication – talking. Effective speech applications do not require any user training they are designed from the end user perspective and think, talk and interact just as any of us would. As recently as five years ago, a number of speech applications were poorly designed and frustrating to use, but technological improvements accompanied by best practices have addressed many, but not all, obstacles and issues. In fact, I must regretfully point out that speech recognition systems are sometimes more compelling than assisted service, as these applications are never nasty or rude, unlike overworked live service agents. If we think back to the 1980s when Automated Teller Machines (ATM) machines (another form of self-service) became popular, push-button banking was often a much better option than dealing with some overworked and harried tellers, at least in New York where I banked.
Speech recognition software yields a rapid return on investment. The payback from a typical speech recognition application is 3 to 12 months, regardless of whether or not a touch-tone IVR is already in place. Speech recognition is not simply IVR with a voice. It is a much friendlier and more flexible interface than touch-tone IVR. Using speech, customers can interact relatively freely, ask questions, obtain information and transact business.
In the past, enterprises could rightfully complain about programming costs and the lack of industry standards. They even had a point about difficult development environments. But no more. As described in my last column, Speech Standards Improve Service Quality, Customer Experience and Reduce Cost, strong and effective industry standards exists and are available at no cost to the enterprise. Additionally, many vendors are now providing user-friendly drag-and-drop development environments for speech (and touch-tone) IVR applications to make it even easier to build speech applications. The new standards have already facilitated the delivery of innovative packaged applications for a variety of verticals. This market is still new and developing, but it is growing rapidly and adding a wide variety of options for system users.
After more than 20-plus years of development, speech technologies are good and continuing to improve, even if true natural language speech recognition software doesnt yet exist for the consumer market. The vast majority are directed speech applications that are natural language-like. The ROI is rapid and has thousands of proof points in companies around the world, including both those that started fresh with speech and those that enhanced an existing touch tone IVR. Speech visual user interface (VUI) experts are available, have an increasing amount of experience and are no more expensive than any highly trained technical resource. And the programming development environments are now very compelling.
The technology is good, customers like it (and occasionally prefer it to assisted service), best practices and resources are available and the ROI is better than most technology investments, so why arent companies jumping to install speech applications? The answer is that enterprises are overly cautious because of the large start-up costs required by speech applications. Given the percentage of failed technology projects and the tough economy that has limited new investments, the concerns are reasonable, but should be overcome by the proven record of success and ROI from speech recognition initiatives.
Today, its rare to see a contact center speech recognition development effort that costs less than $250K and many are in the $500K to $1 million-plus range. Auto-attendant applications will cost less as will smaller contact center implementations, but $250K to $1 million is typical. The start-up costs are substantial, even though packaged applications, standards and new development platforms are reducing the expenses for many companies. Large enterprise initiatives are going to cost a significant amount of money, take 6 to 12 months to implement and involve a variety of internal and external resources all of which adds up to risk. There is no silver bullet to eliminate the uncertainty, but its time to put aside much of the concern. When benefits far exceed the risks, as is the case for speech recognition, risks are worth taking.
The time is right to obtain approval for speech recognition projects, even if you have been turned down by your chief financial officer in the recent past. Build a strong business case based upon a customized ROI analysis and proof points that are similar to your operating environment. Incorporate benchmarks to measure project accomplishments, such as the success rate of first attempts (which should exceed 98% in directed speech applications) and customer satisfaction. Reflect implementation best practices in your business case and specify exactly where the savings are will be coming from. Enterprises, though often risk averse today, are rational and will invest in winning technologies. Its up to you to make the case.