Published on | Prototyping: From UX to Front End — Medium Jordan Harper Credit: The Ironic Giant

We need to develop a modern design framework for digital conversation in order to design effective chatbots.

Gartner predicts that 85% of customer interactions will be managed without a human by 2020 [1] and that shift is not just driven from the business up, but from the customer down: 44% of US consumers say that they’d prefer to chat to bots instead of humans if the experience was right [2].

It’s not a great endorsement of how people view customer service, is it? In fact it’s a pretty damning indictment of the declining quality of brand customer service that it’s become so robotic and scripted that a smart(ish) chatbot can out-perform and out-charm a human.

Chatbots have a number of obvious advantages over humans for customer service:

A million simultaneous agents take up the same physical space and resources as a single conversation (give or take)
A suite of bots can run 24/7, no downtime
Connected bots can search and analyse large datasets instantly to provide relevant information drawn from a multitude of sources
Computers don’t get bored, angry or disdainful (yet)

As artificial intelligence improves, they’re even going to get better at intuiting meaning and dealing with unexpected questions and situations – but until that point, we humans still play an important role.

It’s easy to see how we could end up designing chatbots that fall into all of the same traps that real human customer service falls into — I’d argue that 90% of the 40,000 or so customer service chatbots that already exist suffer this problem — and I’d suggest that designing effective chatbots will need to call on a set of design skills that may seem unfamiliar to traditional digital designers.

Unlike visual interfaces, conversation offers only a linear interaction, so to create a satisfying chatbot experience requires the designer to think more like a writer or storyteller, crafting beginnings, middles and ends for their experiences.

At the same time, it also requires them to think like a developer: dissecting the essential components and actions of a chat and building adaptive decision trees that can appear to ‘react’ in real-time.

It may be that to create effective chatbots we need to define and work to a wholly new set of principles — avoiding the conversational equivalent of uncanny valley by not trying to recreate humans as machines, but at the same time exceeding low-expectations of simply talking to a dumb robot.

In (visual) digital design, we went from being constrained by the medium — pixel fonts, restricted colour palettes — before drifting into the nostalgic horrors of overenthusiastic skeuomorphism. Then came ‘flat’ design like Microsoft’s Metro, which felt ‘digital’ but also cold and critically lacking personality.

Google’s Material Design framework for interactive interfaces

With Material Design, Google have created an ever-evolving, authentically digital but human-centric framework for interactive design – and I believe that finding the chat equivalent of Material Design may be key to maximising the medium.

Some things to consider:

An authentic voice: what should a digital service sound like? Not trying to ‘humanise’ it with pre-scripted jokes and bon-mots — or trying to pass a turing test — but at the same time not appearing like uncaring, artificial robotic echo chambers?
Digitally native communication: when humans talk to other humans online, we rarely write in full sentences. Emoji, acronyms and text-speak have all evolved as native ways of communicating online, and it’s probably more important that chatbots understand these digitally native methods of communication than expending our effort on getting machines to learn how to construct perfect sentences.
Make it flexible: When every ‘agent’ has instant access to vast quantities of information, there is no need for tedious, scripted interactions or referrals to other experts, harnessing this data and being able to branch conversations when necessary will be a key part to making chatbots appear not just a satisfactory replacement for humans, but a better experience altogether.

Thanks for taking the time to read my rambling thoughts. If you enjoyed it, please recommend, share, hit me up on twitter or respond if you want to keep the conversation going!

Designing a chatbot was originally published in Prototyping: From UX to Front End on Medium, where people are continuing the conversation by highlighting and responding to this story.

Designing a chatbot

Cool stuff, every week

More Articles