Chapter 12. Maximizing Efficiency and Clarity

There is substantial evidence, from surveys and other forms of user feedback, that callers care a great deal about efficiency. Efficiency is the ability to complete transactions quickly without wasted steps or wasted time.

Given the serial nature of speech, efficient design of VUIs can be challenging. Consider a system that, in response to a search request, reads back a list of movies playing at local theaters. You can quickly scan such a list visually on a screen, focusing on the columns and items of interest, but listening to spoken playback of a long list of such items can be tedious. Design of an efficient spoken version may require flexible approaches that allow the caller to navigate the list quickly by various criteria, such as "playing after 7 p.m." or "playing in Palo Alto."

On the other hand, VUIs can take advantage of callers' spoken language capabilities in order to increase efficiency. People are adept at structuring complex sentences that package lots of information (e.g., "I wanna go from San Francisco to Boston next Tuesday, arriving by nine p.m."). Provided that the technology can handle such complex input with high accuracy, callers can often express their needs more efficiently with speech than with keyboard and mouse.

Clarity is a caller's understanding, at each point in an interaction, of what exactly is happening, which options are currently available, and how to accomplish his or her goals. Clarity is a product of many design elements, from the wording of individual prompts to the structuring of dialog strategies. Although some of the new natural language understanding technologies provide a great opportunity to better meet callers' needs, they also present new challenges in making the system's capabilities clear to the caller and easy to exploit.

Sometimes designers must make trade-offs between efficiency and clarity. Especially when systems are used repeatedly by the same callers, there may be significant differences between the needs of first-time callers and experienced callers. A number of techniques can be used to accommodate both types of users.

This chapter covers approaches for maximizing efficiency, maximizing clarity, and managing the trade-off between efficiency and clarity when necessary. We also discuss creating appropriate mental models for users of advanced natural language understanding systems.