Skip to content
Why better design leads to better AI
 — April 20, 2022

When I was invited to be a judge for the newly-minted Mobile UX Awards in 2017, I got to see and evaluate many different kinds of mobile experiences — including chatbots. Back in those days, around 10 years out from the launch of the game-changing App Store, UX designers were programmed to apply app design thinking to just about every mobile experience.

But even back then, approaching a chatbot like an app was a mistake. The truth is that AI chat bots and apps simply aren’t created equal, and Frankenstein-ing your way to a product will never be a winning formula, so it was critical for UX designers to develop a unique set of design principles for the new conversational paradigm.

Let’s take a look at the design lessons learned back in the early days of conversational AI.


Bots are a different kind of app

When it comes to user experience, three core components set bots apart from traditional apps: interface, accessibility, and personalization.

With chatbots, conversation (e.g., language) is the interface — whereas with apps, graphical icons and point-and-click menus set the parameters of the experience.

Secondly, chatbots are instantly accessible via messaging apps: No download is required. With apps, users must go through a complete onboarding process before they can access the functionalities they crave. These steps usually involve downloading the app, creating an account, and walking through a tutorial video that educates them about the app’s design and functionalities.

Finally, there’s personalization. Bots offer a 1:1 channel that is designed to create more personalized experiences, whereas apps are usually designed as a one-size-fits-all experience built on a persona’s common denominators.

Rules of thumb

Chatbot design is a different beast, but we are already witnessing a set of best practices emerging from the hits and misses of early days. Based on our experience at Heyday, here are the top things to consider when designing a kickass conversational experience:

Talk like a real person, but make it clear you’re not.

As Paul Graham says, “write like you talk.” Would you ever say “such as” over “like” in a conversation? Probably not, because that’s what you wrote in your college essays to hit word counts.

There’s an easy test for this: Read what you wrote out loud. Does it sound casual? Would you ever say it to your friend? If your answers are yes, you’re on the right track.

Additionally, you should never try to pretend that your AI chatbot is actually a real person. Manage the interaction with integrity from the start — introduce the bot, and users will usually respect its limitations and capabilities.

Make it visual

Just because you’re designing a chatbot doesn’t mean you have to rely solely on your words. You can work in GIFs, emojis, and images to show personality, make someone laugh, or convey options.

Studies have shown that people engage more with visual content. Sometimes you just can’t get past that old cliché that a picture’s worth a thousand words.

Keep it short and sweet

Long chains of text are hard to read, especially in messaging apps where you’re used to quick messages. Think about the last time you saw a three-paragraph text come in from your best friend — there’s just something about it that makes us audibly sigh and feel apprehensive.

Chatbot users are the same way. They just want to get the essentials and get things done. Stick to the message, don’t ramble, and keep it simple. The goal here is to sound human, but with the speed and efficiency we’ve come to expect from technology.


Be clear and specific

A chatbot isn’t like a customer service rep — your user can’t necessarily interrupt or ask for clarification.

The way to work around this and minimize any possible confusion is to keep messages clean and clear. Don’t confuse your user with multiple call-to-actions or data requests in a row. One message should have exactly one intention. And try to avoid open-ended questions at all costs. Err on the side of specificity.

Prepare for edge cases

Don’t leave any dead ends in your chatbot scripts. Think through possible scenarios and make sure there’s a message or a plan to deal with each one. Nothing kills engagement more quickly than a jarring error message. Adaptivity is the key to success in the chatbot space.

Empathy is at the heart of great conversations

As conversation slowly becomes the main way we interact with technology, language — rather than icons or other visual design — will drive the experience.

Although it’s still challenging for designers and writers to create delightful experiences through text and speech alone, there’s a tremendous opportunity for businesses to engage customers on a personal level. After all, what is AI good for if the final experience feels inhuman and impersonal?

For today’s generation of AI designers, the challenge will be this: teaching machines to chat like humans. Humanity and a profound sense of empathy are any designer’s best friend. And whether you’re creating bots or apps, that remains universally true.

Stay ahead of the curve. Get Conversational AI insights delivered straight to your inbox.