In our four-part blog series "How to feed your Chatbot", we highlight the topics of chatbots and voice assistants. In doing so, we would like to demonstrate how such assistants work in general terms and, of course, whether (and how) such assistants could be used for both service and user information.
Before you can breathe life into a voice assistant, you should consider the use cases: How can a voice assistant really help the users or colleagues from the Service department? A voice assistant that is as universal as possible will probably still disappoint its users with answers to misinterpreted questions. The more clearly the field of application is defined, the better the voice assistant can play to its strengths. So a concept must be developed.
In the first blog post of this series, I've already mentioned that a voice assistant who reads instructions for action aloud probably doesn't offer much added value. The context of the read text remains unclear, the reading speed may be too fast or too slow depending on the situation, and the listener lacks the definition of terms that may be specific to the manual. It won't work without pictures.
Define Use Cases
But what do realistic use cases look like? In order to accomplish this, everyone involved has to sit down at the same table: the people who provide information (typically from the Technical Editing and Design departments), as well as the people who will really ask the voice assistant questions afterwards. The latter can be from the service or even customer representatives. Together, we have to identify the current challenges and then develop a user-centered information concept. One way to accomplish this is, for example, with a Workshop that is based upon the concept of Design Thinking for industrial Services.
This makes it quite easy to find out where the individual key use case lies, which inspires customers and employees, and satisfies their need for information both quickly and purposefully. In some cases, concepts in the form of Frequently Asked Questions (FAQs) will suffice to supplement first-level support. In other cases, more complex scenarios such as guided error analysis and troubleshooting may be used.
Target Audience Analysis and Training
Once the use cases are in place, a well-founded target audience analysis is required. Because in order to build a voice assistant, you have to anticipate what users are most likely to ask. And you need a fairly precise idea, because the voice assistant must be "trained" with possible questions so that it can provide the appropriate answers. The training also includes sorting out which synonyms are used for the terms that have been laboriously defined in the documentation. Users should also receive a response even if they don't use the correct nomenclature from the documentation. Depending on the target audience, various methods are available. One possibility could be the Wizard of Oz Method, but you could use it to write your own blog article.
If the target audience has also been determined, then you can begin and the voice assistant can be set up. One way this could look like is described in the third post of our four-part blog series "How to feed your Chatbot".