Blog: What is Conversation Design?
An introduction for lateral thinkers and newcomers in that field.
This week, I was asked by an engineer to define what Conversation Design is and what it means to design conversations. I tried to explain in a way that I hope you can relate to if you come from a different field.
Conversation Design is a design discipline derived from UX design and copywriting. It enables us to use applications and software without the traditional means of a keyboard, mouse and graphical user interface. Instead, we interact by simply using our language. That means while other methods require us to navigate a website or search Google with our hands by typing words or clicking the mouse, with a conversational interface we can just speak. You may have heard of Voice Assistants — you can tell an assistant to turn on the lights just by saying:
“Assistant, turn on the lights please.”
And the assistant will turn on the lights.
Now, this was a very short interaction which did not consist of at least a mutual exchange of language, so we would not even call it a dialogue. You said just one thing and the software reacted to it and completed your command. However, the industry has discovered that natural language can be used to enable customers to interact with these products on even more complex topics.
This implies that the user is not just giving one command. But actually, communicating (=exchanging information) in several turns with the software, which also responds in human language. This can be, for example English, written or spoken.
These days, the interaction between the user and the software increasingly resembles an interaction between two humans which we simply call dialogue.
Within this dialogue, they not only exchange one command but often a few and not just commands. You can see here person A and B have a dialogue over 6 turns.
A: Could you please get me a coffee? */(command from human to human)
B: Yes sure, I can get you a coffee. */(confirmation of the command)
A: Could you give me money for it? */(notification of a condition that is a prerequisite to get coffee for A., and request to fulfil the condition)
A: Here it is. 2 Euro should do. */(completion of a condition)
B: Oh wait, do you want it black or with milk? */(options on how to fulfil the command)
A: Black like Alice Cooper’s mascara.
Enabling this kind of complex dialogue between a human and a machine is what a Conversation Designer does.
How does a Conversation Designer work?
In order to enable this kind of interaction between a human and a machine, a Conversation Designer needs to know what the topic or domain of the dialogue will be. Do you want to order a coffee, just have a little chit chat or — do you want to help people to retrieve their password, support them in using an electronic device or advise them on shopping and help order a new pair of sneakers?
Once the Conversation Designer knows the domain, they can picture scenarios of what customers might want to know or do. Picturing these scenarios is what we also call “Interaction Design” or “User Experience” design. It’s like having an empty space and giving it a ground for people to walk on and directions to help them get where they want — or get to the information they want. In many ways, this work resembles the work of an Architect. In fact, we need to establish a lot of this “ground” and make the space passable, similar to what an architect does and also with a certain purpose in mind. It’s just that our product won’t be physically experienceable. But mentally. In fact, common ground is a linguistic term defined in the early 80ties describing two parties of a conversation making sure they are on the same page.
The bricks we build our architecture with are words. Which are also merely abstract symbols that point at things or events in reality, following the laws of convenience and grammar to have meaning. This is why writers have an immediate advantage when designing conversations as they are used to building that kind of mental architecture. Every piece of text, even this one, is a mental architecture representing things, events and relationships. When a Conversation Designer builds an interaction they do the same thing a writer does — only now they must think of two instances (human — machine). You might be thinking: so, then, this is like writing a dialogue between person A and B. — Almost, we’re getting there.
As a Conversation Designer, writing a few dialogues can be a good way to start visualising the outcome. But if you only consider a single user’s phrasing of a query to to be build into the software, your interaction will rely on the user saying exactly what you wrote or else it will fail. Humans are not that predictable and they always say what’s on their mind. We can anticipate this to a certain degree when they are in a certain situation. When you enter an online grocery store, we can assume that you will want to buy groceries. If you click on the category ‘beverages’, we can assume, you are interested in either alcoholic or non-alcoholic ones and so on. A Conversation Designer needs to break down the situations to find out what a user will want. The more precisely you describe the situation, the easier it is to write what the user would say at that moment. We also need to think of alternative phrases like “I want to get a coffee”, “Could you get me a coffee please?”, or “Do you have coffee?”.
Next, we design or actually write an answer from the bot that would fit these requests. This process has to be undergone no matter how you want to apply this conversational software later — be it as a robot, a non-visual voice assistant with a microphone and speaker or a chat widget.
At some point, I am sure even the writing or speech production will be done by machines, but whatever it is — be it a machine in say 10 years or a human Conversation Designer that designs the conversations, they will all design an interactional structure that enables different dialogues, not just one dialogue.
The Rationale behind Twyla
You will read and hear a lot more definitions and perspectives on what Conversation Design is, often depending on the field the Conversation Designer is coming from. At the moment, many Conversation Designers have an engineering background but this is mostly because this particular area of design is so nascent and still largely tech driven, often leaving creativity by the wayside.
It’s not straightforward to build conversational interfaces, so the bar has been set quite high in terms of those who can access them. This is why so many of those who also have a great contribution to make — such as copywriters, improvisation artists, humanists — are not represented yet in the field. With the rise of conversational interfaces, we need to aim at a more interdisciplinary exchange between engineering and the creative industries, technological know-how and the soft-skilled, human component.
At Twyla, we enable people who have no coding skills to design conversations. If you think about it — conversations are actually a software of their own. Twyla gives people who are not even close to programming or engineering access to the creation of software that can do complex tasks and interact with humans. This is quite exceptional and has huge potential.
At the moment you can create conversational interfaces that can be read and written to: chatbots. This is mostly because this is what our stakeholders are asking for and we have more businesses where a chatbot is more useful than a voice assistant. But who knows what the future will bring…