This guest blog post was written by Sean D. Williams Ph.D., Professor and Director of the Technical Communication and Information Design program at the University of Colorado-Colorado Springs, one of a handful of stand-alone technical communication degree programs in the United States. 

Good conversations: we all know when we have had one and we all know when one went off the rails. Few things are more human than simply talking to one another, and humans have developed extraordinary skill in understanding the most subtle cues of conversation. Human conversation is natural for most people (even introverts), and we don’t really have to think about how it works. It just does. 

Of course, it’s not really that simple, and lots of people have studied conversation demonstrating that it’s quite complex. Fortunately, conversation follows some predictable rules and that has enabled conversation designers to build conversational interfaces like chatbots, machine translation, voice assistants, and virtual assistants that follow those rules to provide personalized and timely support to real people in something that approaches natural language.  

What is Conversational Design? 

Conversation design is all about figuring out the goal of the conversation, the expected experience, and how people will interact with a conversational interface before it even exists. This could be the verbal prompts they will likely give, or the pain points they will likely inquire about from your conversational AI. It's important to get AI in learning and development right whether you're in charge of a product, leading a design team, or developing the technology. If you want to make the most out of conversational AI, it's worth learning about the design process and the special obstacles that come with it. 

Conversation design is particularly relevant in technical communication because the field seeks to communicate clearly about complex subjects in ways that diverse audiences find effective. In technical communication, conversational interfaces are becoming increasingly important because they can deliver technical information in a more natural and intuitive way that increases accessibility for audiences without a technical background.  

Instead of reading through a long technical document, for example, users can simply ask a question and receive an immediate response. This can save users time and improve their experience, which is the goal of a conversational designer. 

Before we dive into some suggestions for effective conversation design for technical communication using verbal user interfaces (VUIs), let’s take a step back to understand how natural conversation works. That background can help explain just why some VUIs work, and some don’t (and why some conversations are good and some aren’t). 

How Does Natural Conversation or Human Conversation Work? 

In the field of conversation analysis, a couple key ideas explain most interactions: “adjacency pairs” and “pragmatics.” 

Adjacency pairs is the simpler concept and says that successful conversations normally occur in two parts that rely on taking turns. For example, if you and I are talking in the hallway, and I invite you to my birthday party, you almost certainly would answer yes, no, or maybe. The invitation (part 1 of the adjacency pair) demands part 2, your response to my invitation. If you didn’t respond (or changed the subject) the pair is broken and the conversation fails.  

Placed in context of VUIs, if you ask a chatbot or voice assistant to provide you with the current speed of your Internet connection and the VUI doesn’t reply, then you assume it's “not listening” and you become frustrated. If it does reply, then you had a pleasant “conversation” with the AI chatbot. Your needs have been met in a way that works like a natural conversation where you asked a question and received a satisfying answer. 

The number of possible adjacency pairs is quite extensive, but the key lesson is that conversation occurs in pairs where the first utterance requires a fitting response.  

Pragmatics, created by a language philosopher named John Grice, is more complex and provides four rules (or “maxims”) of conversation:  

  1. Maxim of quantity: Speakers should provide as much information as is necessary, and no more. 
    If you ask a chatbot for the weather, the chatbot should provide the current temperature, as well as the high and low temperatures for the day. However, the chatbot should not provide a long, detailed explanation of how weather patterns work. By sticking to the maxim of quantity, the chatbot provides you with the information you need, without overwhelming you with extraneous details. 
  2. Maxim of quality: Speakers should only say things that they believe to be true. 
    If you ask an AI chatbot for the location of a nearby restaurant, the chatbot should only provide information that is accurate and up to date. If the restaurant has closed or moved to a new location, the chatbot should not provide incorrect information. By sticking to the maxim of quality, the chatbot maintains your trust. 
  3. Maxim of relevance: Speakers should only say things that are relevant to the conversation. 
    If you ask a chatbot for information about a particular product, the chatbot should provide information that is relevant to that specific item. This might include product features, pricing information, or customer reviews. By sticking to the maxim of relevance, the chatbot provides you with information that is useful and applicable to your situation. 
  4. Maxim of manner: Speakers should use clear, concise language that is easy to understand. 
    If you ask a chatbot for help with a recipe, the chatbot should provide clear, step-by-step instructions that are easy to follow. By sticking to the maxim of manner, the chatbot ensures that you can complete the cooking task successfully. 

Adjacency Pairs and Pragmatics represent just two approaches to conversation analysis, but these two might be the most important for understanding the basics of how conversation works. Equally important, the concepts are easy to comprehend and therefore relatively easy to consider as you design a conversational interface.  

What Are Some Tips for Implementing These “Rules” into Conversation Design? 

Given these “rules” of conversation, here’s a few tips that you might consider that draw on skills and practices already familiar to technical communicators: 

  1. Know your audience. Before you begin designing the conversation, understand who your audience is and what user experience they expect. What are their needs, preferences, and pain points? What kind of language do they use? By knowing your audience, you can tailor the conversation to their specific needs and create a more effective user experience.  
  2. Use natural language. When writing for the interface, use language that sounds like a human speaking. Avoid using technical jargon or overly formal language that might be difficult for users to understand. Instead, use simple and concise language that is easy to follow. 
  3. Provide clear options. Provide clear options for users to choose from that guide the chatbot conversation and ensure that users can find the information they need quickly and easily. Providing clear options means first creating a knowledge base of common questions and answers to guide the conversational flow so that users and the interface share a common context. 
  4. Test and iterate. As with any design process, you should collect data on usage from real users. This can help you to identify any issues that might exist in your interface, revealing ways that you can iterate on your design to improve the user experience.  

Why Integrate Conversational Design into Technical Communication?  

While building conversational interfaces can be somewhat challenging, VUIs do have several benefits that reflect technical communication’s focus on helping people solve their problems:  

  1. Improved accessibility: Conversational interfaces can make technical information more accessible to users who might not have a technical background. This can help to improve user experience and reduce frustration. 
  2. Personalization: Conversational interfaces can be customized to meet the specific needs of individual users. This can help to improve user engagement and satisfaction. 
  3. Immediate support: Conversational interfaces provide immediate support to users. This can help to reduce the need for users to contact customer support or seek assistance elsewhere. 
  4. Improved documentation: By analyzing user queries and interactions, technical communicators can identify gaps in their documentation and update it accordingly. This can help to ensure that the documentation is current and relevant to users. 
  5. Cost-reduction: Conversational interfaces can help to reduce the cost of providing support to users. By automating support, businesses can reduce the need for human support agents. 

Conversation design is a powerful tool for enhancing user experience in technical communication. By incorporating conversational elements, such as chatbots or voice assistants, you can present technical information in a more engaging and user-friendly manner that comes close to the natural conversations that we already know how to have. If you’re not using conversation design yet, I encourage you to give it a try!