Let’s talk about pauses. Pauses determine how fast or slow your system will be perceived. One thing to keep in mind is that people in your business who are familiar with the system, such as your testers, or your business unit, will rapidly learn how to use it and become power users. They are insiders.
So, unless your callers are also power users, which means they call at least once a week, they will not have the same experience of the system that your internal users have. One side effect of this is speed. Your internal users will perceive a system as being “too slow” because they’ve listened to it dozens if not hundreds of times before. So, if your system “seems slow,” take this with grain of salt. It’s usually better to err on the slow side. So if you hear this complaint, run it by a naïve user before you make efforts to speed it up.
Pauses are the “white space” of your VUI design. They occur in three different places:
During menus, between items
At the end of a menu to allow for an additional option. “Or, say ‘It’s something else’”
Between menu repetitions (No input timeouts)
The pauses between items in a menu is a pause which will allow your caller the chance to think about that choice and either remember it, or reject it. This is a very important pause since it contributes to the comprehensibility of that menu. This pause should almost always be longer than the natural pause your voice talent will create during a recording session, which are usually around 250 milliseconds. Instead, have your audio engineer cut in a pause of anywhere from 500 to 750 milliseconds.
The pause at the end of the menu to allow for an additional item needs to be long. It’s designed to allow the caller who hears what they want to barge-in over the prompt. (Make sure to enable barge-in when you do this.) But if it’s too long, there may be a turntaking issue, where the caller starts talking, then the system starts talking, and so on. This pause should be at least 1-2 seconds. Maybe longer. How long, exactly, is going to depend on your caller base, and what that prompt is. Start with 2 seconds and keep an eye on it in tuning.
The pause between menu repetitions, or the no input timeouts, come with a VoiceXML default of 7 seconds. Some platforms have a shorter default. Anywhere from 4 to 7 seconds seems to work, but longer will create a slower feel and also give the caller more opportunity to speak. Go longer if you have a caller base who may be a bit slower to react to voice prompts, such as second-language speakers, or an older population. As always, monitor your system and make sure your user interface fits your callers’ needs!
This was originally published on the blog of my employer, Versay.com.