How do we Let Conversation In?

My good friend and fantastic Product Manager, Ty White, recently forwarded me Benedict Evans’ article, “Chat bots, conversation and AI as an interface.” Evans’ article triggered all sorts of questions as I read. I started to write out my thoughts in an email, but it started feeling way to big for an email. I’ve captured my reflections here, instead — more questions and pontifications than assertions or solutions. I’d love to hear your thoughts, as well, especially if you think you’ve figured some of this out.

On Benedict Evans’ recent article on chatbots and conversational UIs, he wonders:

“Instead of going through all the pain of persuading someone to install your app, what if you could have an immediate, fluid and easy way to engage through a conversational interface?”

This has long been a problem for apps and, more broadly, for services in general. Shortly before reading Evans’ article, I had spotted a 2015 quote from Elias Roman (following the Songza acquisition by Google and the introduction to free listening in Google Music). Roman was commenting on asking for signups and payments before even giving a person a taste of the service on offer:

“We’re basically asking for people’s phone number before having a chance to flirt. The single biggest door we can open is a free tier that lets people try the experience.”

Both Roman and Evans (along with so many others) are looking to lower the barriers between a product and its potential users, in this case with particular regard to customer acquisition. Of course, with conversational UIs that operate through SMS, the product is literally asking for a person’s phone number before even asking their name — flirting be damned!

Sometimes, growth and adoption is all about lowering barriers and reducing friction. An example of this that has become nearly ubiquitous is social login. Allowing sign-up and login via Facebook/Twitter/Google has virtually eliminated a whole slew of repetitive user inputs and often provides a richer identity for a person to use on the new platform. With conversational UIs, removing the obligation to download an app or even open a web browser seems like a natural progression toward zero-friction acquisition. I think Evans is on the right track in this respect.

On the other hand, there’s plenty of evidence that it can sometimes be better to retain a little friction upstream in order to provide a better downstream experience.[1] For instance, asking critical information during new user on-boarding may add a step or two, but whatever abandonment you see in your new user acquisition funnel may be offset by higher-quality user engagement, or a reduction in customer support contacts or churn a day, a month, or a year later. I wonder if this same thinking applies to that initial flirtation, as well. That moment of introduction where the product tells a person what it is and how it can help them. Could this be done through a conversational UI? Sure, probably. But is SMS the best medium for that sort of interaction? I’m not so sure.

I think Evans is totally on-point with his observations on conversational UI: this isn’t something we should pile onto our existing interfaces, but rather an opportunity to provide a different type of interface for the people who use our products. It’s not an interface that will work for all types of tasks — at least not yet. But it could be a type of interface that could lower the barriers for users who have a hard time justifying another app download, especially when they might only have a few hundred megabytes of storage left on their phones, anyway. (Ugh!)

Evans draws toward conclusion with the following:

“But it doesn’t follow that that’s a great way to do everything. If it takes you more time to work out what you can ask the AI assistant than to drag the meeting to a new slot on your calendar, you’re doing it wrong. An AI shouldn’t be more mental load than just tapping the damn button.”

I love the opportunities that conversation UIs and chatbots open up for us, and am excited that commercial AI technology has finally made it this far. Not long ago, I commented on an excellent article by Tony Aubé that there is a good reason most of our user interfaces are graphical, but that it is a shame our products don’t make better use of haptic, auditory, or other types of interfaces. It’s like having a toolbox full of great tools, but only ever using the hammer. Ultimately, whether we’re talking chatbots or haptic feedback, it will be the responsibility of we designers and creators to ensure that we pick the right tool for the task at hand.