Speech technology: great gain for just a little pain.

I hate to use the word "holistically" (or its even cheesier, more new-age spelling, "wholistically") since it was one of the most overused words of the 1990s, referring to everything from business processes to herbal supplements to snack foods, but it applies to the call center industry so, well, wholly, that it merits dragging out of retirement. There are areas of the call center that ought to be planned holistically but sometimes aren't (traditional IVR menu design, for instance), areas where process are seldom applied holistically except in the best contact centers (CRM, for instance) and areas where the process must be holistic or it just won't work. Speech technology falls into the latter category. If it's not planned right from start to finish, it will represent the equivalent of lighting up hundreds of thousands, or even millions, of dollars with a butane lighter and watching the money burn. In a speech implementation, if each department has different ideas and requirements and they don't get involved in the multi-step process from start to finish, they will be left out, and the company will find an end result that suits, at best, one department. At worst, the finished product will suit no department at all, and will leave your customers baying for your blood.

This month, we have two excellent articles on speech application development and design. The first was supplied by Christoph Mosing, VP of Professional Services at Envox Worldwide. Writes Mosing, "Many organizations fail to engage the right people and have the right conversations at the kickoff meeting. This results in project managers being shocked later in the project: management doesn't like the chosen persona, data cannot be accessed as specified, etc. To ensure success, continually check that everyone's expectations are in sync, provide major stakeholders with regular updates and get buy-in from all parties as you go. While this may seem obvious, it is easy to skip these steps while in the middle of a project."

The most prophetic words in the call center industry are, "While this may seem obvious ..." Immersed in the call center industry as most of us are, armed with information and knowing what we know, it's immensely shocking when we (regularly) run into call centers that continue to make mistakes that are explained in the first chapter of the most basic call center management book.

Do you have a poorly designed billing system? You're bound to hear about it from your customers, but it probably won't keep most of them from doing business with you. Bad CRM? Considering there was no CRM (at least, it wasn't called that) 10 years ago, you might be able to limp by as a third-rate company without it. Bad IVR menu tree design? That's more serious. You'll start to lose customers. Poorly designed and implemented speech solution? Customers? What customers? You won't have any.

Testing is the key. As Mosing of Envox writes, "Many organizations try to make up for project delays by cutting the testing and tuning time, only to end up with more problems down the road." That just may be the understatement of the year.

The complexities of speech and the necessity to get it right have scared a lot--if not most--call centers away from speech. This is a shame, since call center technology follows life: the greater the risk, the greater the reward. A well-done speech implementation in a call center can bring effective and efficient automation that outdoes the automation potential of all other call center technologies put together.

Deanne Harper, Manager of Speech University at Nuance points out that the problem is often that rather than designing a speech application for callers, a faulty system will try and design the callers to fit around the speech application. She writes, "Rushing design decisions without addressing a comprehensive series of design questions increases risk. A good design is caller-centered. It recognizes and addresses caller needs and expectations. It incorporates strategies to optimize recognition accuracy. Without clear requirements defined in advance, a system design will yield confusing or misleading prompts, as well as an inconsistent prompt style and error handling strategy. It won't contain the functionality specified in the requirements. The net result is that users may experience difficulty using the system."

Users may experience difficulty using the system. That's a nice way to say, "Your customers may end up screaming at your system into the phone, pulling their hair out in hanks and gibbering in rage."

Errors in IVR menu trees become rather obvious rather quickly because of the limited number of options presented to callers. With the complexity of speech and the almost limitless way people can express themselves to more advanced natural language processing systems, mistakes won't be so readily apparent, and will certainly make themselves known only after the launch of the application, at which time they must be corrected as soon as possible.

Says Nuance's Harper, "As it can be difficult to anticipate and exhaustively test for every potential situation and condition before a speech system goes live, you can expect to make changes. Your organization should have processes in place to identify and rapidly resolve the production issues identified during tuning. This process will minimize your callers' exposure to negative elements in the application."

And thus, minimize your exposure to torrents of obscenity-sprinkled customer abuse and its resultant lost business.

I'm a great believer in the exercise mantra, "No pain ... no pain." But the potential gain that can be realized via speech for a call center is staggering, and today's implementations--some of them via hosted solutions that greatly reduce responsibility and headaches on the call center's part--can help greatly minimize the potential pain.

The author may be contacted at [e-mail].

By Tracey E. Schelmetic

Editorial Director, Customer Inter@ction Solutions

[ILLUSTRATION OMITTED]