Opening the Kimono
Lessons learned from some past speech deployment mistakes
was recently pulled into discussions with several allowing the organization to pick and choose recommenda-
I organizations for whom speech did not meet expec- tions and often rendering each vendor’s advice ineffective.
tations. In each case, the organization’s leaders threw up • Lack of predeployment usability testing—One of the
their hands, cried “uncle,” and went back to their comfort- biggest advantages of usability testing is its predictive power.
able touch-tone applications. Now, some folks are passionate If a test is well-designed and the participant sample matches
about speech technology, so the mere inkling that someone the user population, then you get a pretty accurate idea of
might be less than a fan seems hard to fathom. On the other what to expect when an application deploys. In many cases,
hand, keeping a circumspect, decidedly agnostic view of tech- usability testing wasn’t done, or it wasn’t done well, which
nology is what human factors folks do best. It’s easy to emo- made production usage a giant (and unwelcome) surprise.
tionally react in a so-called “failure” situation and conclude • Spaghetti-like business logic—Although this is a very
that speech technology is useless—a conclusion of conven- common problem, its implications are clear. If the script is
ience that’s not necessarily accurate. hamstrung by business logic that requires a historical
For our New Year’s resolutions, we speech types need to be overview and organizational chart to explain, the user
a little more reflective and honest about the gating factors on won’t be able to navigate the application.
telephone-based interaction success (speech • Lack of change management—This
and touch-tone). Designers can support Designers can problem came in two versions: design,
success by knowing and helping to mitigate support success by deploy, and put it in the closet and forget
the pitfalls that make speech unsuccessful. knowing and helping about it; or let anyone with an opinion pro-
If we peek inside the kimono, we will to mitigate the pose and implement changes. In many
see that each organization where speech pitfalls that make companies, changes were proposed for
failed had some combination of the fol- speech unsuccessful. unverified problems, introducing more seri-
lowing characteristics: ous usability problems than what origi-
• Decentralized control of the interactive voice response (IVR) nally existed. From there, it was a slippery slide to unusability.
system—This didn’t necessarily mean the company didn’t