This post concludes my recap of a three-part podcast series on self-service design with Judith Halperin, principal consultant in speech engineering at Avaya. In two previous posts, I related Judith’s observations on designing a user-centric and user-friendly self-service experience and on capturing the benefits of self-service while providing a superior experience. In this installment, she talks about the role of language in designing more natural, or rather, more human self-service experiences.
Click here to listen to the full conversation.
What does “more customer-friendly” really mean? First and foremost among guiding design principles is to minimize your customers’ effort. For instance, if we’re talking about the caller experience, customer effort could refer to how much information your callers have to retain and use in order to have a self-service interaction, or how much information they have to provide just to get started.
Today though, let’s focus on a commonly broken rule. When you call a company for customer support and the IVR offers you 10 menu items, by the time you get to the last item, can you remember what the first one was? Remember that callers have to keep all of the options in their heads, a much more difficult task over the phone than on a website, a mobile app or via webchat.
So how can you make it a better experience for your customers? First, do some analysis and prioritize options based on popularity. A similar principle applies in website, mobile app and webchat design–you want to keep your most popular choices above the fold. At the same time, limit the options that your customers have to remember: generally five to seven items per menu as a guide. Also, allow your callers to start talking as soon as they hear the option that interests them.
Stay mindful of the recency rule. Again, most people tend to remember best what they heard last. It may be useful for the self-service system to offer a prompt right before the customer is asked to respond. This can help guide your customers to make the right response needed to serve their needs.
It’s also important to factor in the role of language in self-service design. For instance, consider implementing yes or no questions where it makes sense, as they are much easier interactions for callers. Also, if universal commands such as “agent” are used, be consistent; don’t substitute “operator” or some other alternative in a later menu.
Prompt structure matters too. For example, leading with “Which could I help you with?” implies that a list of options will follow. This type of prompt tends to be more straightforward to customers than the more open-ended “What can I help you with.”
Regarding speech recognition, the software-based recognizer can only listen for items that are in its set of established grammars and is predisposed to assume that the caller is going to say something it expects to hear. Also, a recognizer can get only so much information from a person’s voice, so it is important to provide acoustically different options that the recognizer can clearly distinguish between. Avoid such combinations as the classic (well, classic in the world of self-service design) “repeat” and “delete,” which differ only by a consonant sound.
Self-service systems can be powerful tools for businesses. Designing a system with customer-centric, natural language principles can go a long way toward the success of your self-service experience. Do you have some thoughts to share on self-service design? If so, we’d love to hear them.
