The art of interaction design is about understanding human behaviour and how that relates to how we humans engage with technology. It is about creating interfaces and experiences that are based on a sound understanding of technology and human behaviour and applying that knowledge to create intuitive, easy to use designs.
And interaction design principles; that is design standards and rules of thumb based on human psychology, are useful guides for us UX’ers when we are creating new experiences. For example, knowing more about cognitive processes such as visual perception e.g. how we experience and see the world around us in patterns (Gestalt principles), helps us design interface elements such as menus, that speak to that human trait. Interaction design principles and design standards are useful in supporting us to building interfaces that are easier for people to recognise and engage with. However, what happens to our standards and guidelines when our ‘interactions’ turn into conversations?
When the interface is no longer the way to communicate with our users
What happens to our interaction design principles when designing experiences for voice or conversational UIs? When there is no longer an interface to see, Gestalt principles of how we see and group information together visually become far less relevant. Does this mean that our traditional interaction design principles and guidelines are now redundant? The short answer is no, I don’t think so. Because technology may change but human needs don’t. We still need to build experiences based on our knowledge of human behaviour and we still need design principles to guide us. Our traditional interaction principles will, just like technology we use, need to evolve somewhat.
What might those new guidelines look like?
Design considerations for conversational UI’s
Obviously our knowledge will grow as we work and explore more in this space but some thoughts on what could help us get started when designing these experiences:
- Start with a script. As Oren Jacob former CTO of Pixar (via Google on Designing Conversations) says ‘think of this space as interactive screenwriting. We’re responsible for lines one, three, five, and seven, but not two, four, six, and eight. The even numbered lines, of course, belong to the users”. Thinking of the conversation as a script will help us when starting to plan and build our decision trees.
- Take the lead in conversations, make assumptions and learn. To avoid endless clarification input from the user and stilled experiences, make assumptions about the user and learn and adapt the conversation as it builds. For example, we could reasonably assume the user name ‘Charlie’ is male unless the user tells us this is not so.
- Consider how the user ‘locates’ themselves in a conversation. In traditional usability testing Steve Krug tells that some great questions to test the user’s overall understanding of the system is to ask them – Where are you? Where you have been? And Where do you think you could go next?
- We want to avoid users getting lost in conversations, which is especially important in this new medium as we don’t have traditional UX elements such as menus or progress bars to support our user’s understanding. We need to think about how the user could go back a step, or how they might save a conversation and indeed how can we support them in getting an overall sense of where they are in their journey and how long it might take them to complete it.
- Design for Limited memory. Cognitive overload, that is asking the user to remember too many pieces of information, is especially important to avoid in this new medium. We have all had the experience of IVR telephone systems reading out long lists of options which are hard to remember. When designing conversations, considering human biases such as the serial position effect, that is that people tend to remember what is first (the primacy effect) on the list and last (the recency effect) on a list, this will impact how we build the dialogue.
- Dealing with errors and avoiding “this does not compute”. Dealing with errors has always been a key question for us UX’ers. This new medium will present interesting challenges for us to solve if our users make a mistake or change paths midway through our conversations and we need to think how we deal with that in a more human, less machine-like way. With machine learning we have some advantages here in that we can be more flexible and adapt the flow somewhat based on user input and reaction rather than rigidly pushing users down a fixed decision tree or path.
- Design with emotion in mind. Understanding and connecting with our users’ emotions has always been at the heart of good experiences. Chatbots with personality and voice apps like Alexa have the ability to be perceived as more human (users saying thank you to Alexa is not uncommon!). Which means there are even more great opportunities for us UX’ers to forge strong emotional bonds with our customers through this medium.
AI brings with it tremendous opportunities to create personal, relevant and engaging experiences. Not only will it involve machine learning but also human learning, from both the designer and the user, on how to engage with this new medium.
We Ux’ers are not however, starting with a completely blank page, we still need to apply our understanding of human behaviour to create great designs and interaction design principles help us do that. However, we will need to pivot somewhat in terms of how we think about how we apply them to design and in particular how we create visual solutions. When creating experiences for this new medium, we will need to learn about how we express our designs more in terms of words, patterns, characters and unfolding stories.
My previous blog talks more about Designing conversational UI’s using storytelling