Conversation Design and How to Approach It
Conversational design is a combination of several disciplines which include copywriting, design of user experience, interaction design, sometimes visual design and motion design, and, if relevant, voice and audio design.
When you consider conversations between real people, they are immediate and very much dependent on the context. That is why creating such a conversation between the end user and a chatbot takes more insight than might seem at first.
There are three main principles of conversation design:
- Cooperative Principle
- Turn Taking
According to this principle, effective communication among two or more people relies on the premise that there is underlying cooperation between the participants.
Implication in conversations means that what we say in a conversation is hardly ever literal, so we subconsciously identify the real intention of the speaker based on our life experience.
For a conversation to feel natural the cooperative principle needs to respect what is referred to “maxims:”
Maxim of Quantity
Means that the speaker gives the listener only as much information as is necessary to further the purpose of the conversation.
Maxim of Quality
You need to share quality and truthful information.
Maxim of Relevance
Share information that is relevant to the topic of conversation.
Maxim of Manner
Imitate real person-to-person conversations. Even though people are aware they are talking to a chatbot, they still expect it to follow the rules of human conversations.
To make your conversation functional allow for the chatbot and the end user to take turns in the conversation.
One of the roles of the conversation designer is to put transitional prompts (such as questions) into the conversation which need to be explicit but sound natural.
When designing conversations, you have to take into account the end users’ context.
Different types of conversations require different sound of voice which needs to focus on the end user’s physical and emotional context and the kinds of conversations that can serve them in that context to make your users feel comfortable during the conversation.
After considering the principles of conversation design, let's take a look at the conversations elements to keep in mind when designing your conversations.
In real life you would start a new conversation with a greeting and introduce yourself to a new person. It makes sense for the bot to do the same.
Example: Hello! – Welcome! – My name is …. and I am here to…
Finish your conversation with something that sounds like a natural ending to a conversation to provide closure to the end user and highlight the bot’s social intelligence.
Example: All done here. Thank you for contacting us. Goodbye.
Question is one of the most effective prompts to keep the end user engaged and gather the information you need.
Example: What can I help you with today? – What’s your name? – What credit card do you prefer?
Necessary to ensure the end user that their input was received and saved.
Example: Okay. – I understand.
You can also send information to your end users but bear in mind that a prompt should follow each informational statement – you want your end users to keep on engaging with the bot.
Example: Here are the locations of the nearest bank branches. Can I help you with anything else?
Suggestions are also a way to engage your end users and help them with decisions, or hint at new features.
Example: If you are interested in a new credit card, I can show you what you are preapproved for. Do you want to know more?
Apologies are important and necessary but try to avoid using them too often as they highlight the bot’s shortcomings. You should follow the apology with offer of an alternative option.
Example: Unfortunately, I couldn’t find a branch office near your location. Do you want the location of the nearest ATM possibly?
Commands should be camouflaged as conversation hints because you want your end users to realize through the conversation what to do next, you should not have to teach them.
Right: What can I help you with next?
I want to find more about insurance. – I want to talk to an agent.
Wrong: To continue select either “insurance” or “talk to an agent”.
You should provide confirmation to end users, if possible, so they would have assurance that their queries were correctly understood and handled.
Example: You wanted to know the current status of your checking account. Anything else I can help you with?
We all use discourse markers in everyday conversations (words like anyway, right, like, etc.). They aid in comprehension and make the conversation sound more natural.
Examples: By the way – Above all – Fortunately – Right - Anyways – In that case – Ironically – In other words – Consequently – Also
– For that reason…
By the way, did you know I can also help with deciding which insurance is the best for you?
In that case, let me re-schedule your flight and have a safe trip!
What we consider errors here is when there is a break in the communication either because the user didn’t respond, or possibly said something the chatbot can’t process. This can occur at any point in the conversation so it’s good to have a fallback option configured for such cases.
Example: Sorry, there seems to be something wrong with the address you sent for the closest bank? Can you please send it again?
Audio and video element, gifs or images can be used to answer questions and add personality to your chatbot.
This will largely depend on the type of persona you decide your bot will be.