Designing conversations

Facebook just announced Messenger and my feeds are all over it. That got me thinking about it again, because if “it” is happening, I want to be ready.

I am a UX Designer.

I’m not schooled to design for conversation, which is funny, because conversation is at the very core of human interaction. Instead, “my” users are taught to work with a screen and a pointer to get things done and I use these interfaces to design their interactions.

Conversational UI feels like the future, but it also feels like we’re going back in time, to what really matters. The main reason it’s taking off, is that technology is currently in a state where computers can actually have a good enough conversation with us.

But despite of this, everyone seems to agree, that no-one has built a good enough conversation yet.

But if it is really taking off as everyone seems to agree upon, how the hell would I design it. How does one design a chatbot?

For argument’s sake, I stick to a Facebook Messenger chatbot and not any other interface, because, as some sources say, Messenger could well mean the death of the App Store…

How can I make it valuable?

It can only be valuable, when I’m able to offer something that is (significantly) faster and simpler than the alternative. Some of the examples that Facebook uses to promote the new Messenger service are just plain stupid: Starting a conversation with a Messenger weather app takes way longer than simply typing “weather” in Google.

And this is just one of the bad examples that shows that so far it’s “only” a technology. Which is great, but as with any new technology, it will only succeed, if we understand it and use it to create something valuable for our customers.

So, understanding my customers will remain relevant to design a conversation — which means I’d still have a job.. But it goes further than the conversation itself. What is the value in chatting with a brand, instead of going to their website, or using Google to find it?

The need for a conversation has never been on my list of customer needs before, so why should it be now. Or is this new technology a means to solve any complexity that couldn’t be solved before? Can’t I solve that by changing the IA? Or is it an interface to make a brand feel more human? But don’t they notice it’s a bot? Does it matter?

Defining the value proposition for a conversational UI is crucial. And probably the hardest part, is to convey that value proposition to users and as you can see here, KLM needs to give away tickets to do that.

How would I design the conversation (not the interface)?

Should I approach the same as my current design challenges, or would I need a more systematic approach to crafting these conversations? It will be pretty complex, with lots of variables, states, routes, possible questions and answers. At first glance, it looks infinite and very mathematical. And that’s probably why we only see crap examples.

There are no books or resources that would help me to design a conversation, or at least, I haven’t found it yet. There are hundreds of books written about designing for screens though.

There are no design patterns or guidelines to design for conversation. Screens suddenly seem easy.

What would the deliverable be, to show the answers my bot might have. Will it be a really long copy doc, or a “conversation blueprint”? Is it even possible to create a deliverable?

How would I research what my users will ask? Would I need to anticipate all questions any user might ever have? And how about different languages?

Designing for conversation seems like a different discipline that has nothing to do with UI design, so I would probably need to use different tools too and work with different people. Visual designers wouldn’t know what to do I reckon.

How can I ensure quality?

My boss, David Vogel mentions this intangible factor “quality” in one of his talks. The sum is greater than it’s parts. All designers strive for quality, but how can I ensure quality, when I design a conversation?

It’s possible to test if it WORKS good, and Facebook made sure it LOOKS good. Quality is when something FEELS good. It’s hard to put in words and it’s even harder to test it. You just know it when it’s there.

Will there be a moment where I go: “Damn-shit-yeah, this is a great conversation”.

I hope I will.

Can I give “it” an identity?

When I think about conversational UI for brands, they will most likely look a lot like each other. The conversation I would have as a user with Zara would probably be similar to the one I would have with H&M, or with Primark, or Louis Vuitton. But they’re very different brands.

Would it be possible to add a brand’s identity to a conversation? It would definitely be possible in the tone of voice, but other than that, this new design space feels very limited.

But then again, the mobile space felt like that too at first.

I have thousands of questions and I’m not sure if I like this limited working space to be honest, but I can’t wait to design my first conversation.

When I have the answers, I’ll let you know.

Some good readings:

http://alistapart.com/article/designing-the-conversational-ui
http://alistapart.com/article/all-talk-and-no-buttons-the-conversational-ui
http://www.fastcodesign.com/3058818/how-facebooks-big-bet-on-chatbots-might-remake-the-ux-of-the-web

and someone else with the same problems ;) 
https://medium.com/@beerud/conversational-interface-design-the-new-ux-frontier-3be054e7302a#.pa1jeywon