Making a conversation UI? Read this first.
As we approach the era of bots and AI in a broad consumer context, a convention for conversation is going to become increasingly important.
Conversation UI vs. Chat UI
Conversation UI is one of two distinct UI-approaches to bot communication.
Conversation UI implies natural language parsing and conversation.
In my humble opinion, we are not ready for conversation UI, unless you enjoy artificial conversation. Bottom line: artificial conversation is annoying and largely time-consuming.
Chat UI on the other hand is an interplay of text-triggers, yes-no-maybe buttons, directives, and reactions. Responses from the bot can be rich but should not obligate conversation beyond yes-no-maybe.
This distinction is the first important consideration in building your bot’s artificial personality, but remember: conversing with a bot is not the goal.
The purpose of a bot is to be easier and faster than a traditional interface.
But wait. There’s more.
Because a bot communicates through a messaging client, it still needs to relate to its user through language, and this affords the designer a new kind of creativity. It’s not visual, but it’s no less impactful design-wise and needs to be done thoughtfully.
The Content Layer
This is the stuff you need to communicate. It’s the what.
“Hey Yala”
Yala says, “I’m listening, Gary”
Bot content is quite fragmented. Neatly organizing these fragments require classes and tags. A class for the Yala response above could be called, “aye” and a tag could be something like “cheeky”.
Now you have the how and the what.
The Tone Layer
Once you have your content, and you can identify your content fragments in an organized and semantically articulate way, you’re ready for tone. Is your tone figurative or is it to-the-point? Is it friendly or cheeky? Tagging your content-fragments appropriately will help you provide platform-specific tones. SMS conversation should be more to-the-point, whereas Slack conversation may feel better with flair.
The Idiom Layer
The idiom layer is culture-specific, gender-specific and geography-specific. Nuance, connection, irony and interest are all contained in this layer. In my own conversation design I haven’t quite gotten here yet, but it’s interesting to think about.
Idioms are extras that can (and should) inform the way your content and tone are expressed. Idioms should be applied to generic responses and be informed by some information you can easily gain from the user.
Do you have anything to add to this? Please share comments below.
Favorite this article and sign up for Yala to witness her quantum-machine-learning smarts in publishing your social media posts to the right network at the right time.
