How to give your product more life - Mixpanel
Blog Post

How to give your product more life

Parker Tarun

There isn’t a lede about Chris Messina that doesn’t mention his most famous invention, the hashtag. I’m not sure whether this irritates him or not. What I do know is that hashtags facilitate conversations on the internet, and that’s integral to Chris’s ethos: He likes to start big conversations.

Fittingly, Chris has made a reputation for himself as one of the leading voices in the conversational product space — an area he loosely defines as products that use “messaging and voice computing.” This includes chatbots, voice UX, and quite possibly much more. Chris’s finger is on the pulse of one of the most cutting-edge and misunderstood technologies out there.

Having spent time at Google and Uber, Chris has seen the good, bad, and the ugly of how cutting-edge companies start conversations (or don’t.)

Messenger platforms and voice are more than a delightful add-on to existing products. They’re a whole new category. To build them successfully, product people need to rethink their approach. Even conventional product builders, who aren’t sold on any of this, can learn from their example.

“Conversational products relate to delivering service over voice and messaging interfaces, yes,” Chris says. “But it also implies a different philosophy when it comes to product development, which is constant iteration—constant A/B testing.”

Most PMs would say their cycles are already in constant iteration. But can they iterate at the rate of their users’ speech?

It was hard to build Alexa. It took Jeff Bezos cornering an ailing industry (see: and creating a cash-cow out of a burgeoning industry (see: AWS), before he was able to throw resources at a voice command-powered product that could play The Weeknd at Chris’s beck and call.

But creative product managers and designers have as much a chance in this economy as Bezos. Conversational products aren’t for the rich. They’re for the fast and the versatile. And even if you don’t care to build one, you have a lot to learn from those who do. 

Design discipline

At the beginning of 2016, Chris joined Uber as Developer Experience Lead. He’d helped grow the developer ecosystem at Google, and now he was going to do the same for the Uber Developer API.

Conversational contexts and Uber? They seemed like a natural fit.

“My strategy was to demonstrate how Uber enhances other people’s apps by taking care of this logistical question of how do I get from point A to point B,” Chris says. “Conversational products, in particular, are well-suited to those types of integrations, especially if they provide recommendations rooted in the physical world.”

Say you have a concert app with a conversational interface. Is it so absurd to think a user might ask your app how to get to the show?

But if all these different apps were going to call the Uber API, Chris wanted some consistency. His background is in design, after all. If calling a ride looked different across all these apps, that’d be confusing for users.

“The ways in which you need to invoke Uber through messaging or voice had to be consistent,” Chris says. “You should be able to go deeper and deeper into accessing services provided by Uber through conversational interfaces. To do that well, we needed to actually develop a design discipline around those challenges.”

For bots to be successful, maybe Uber needed to hire different kinds of talent, like people who focused on character development: improv comics, writers, script designers. Chris saw an opportunity to start a bigger conversation about how users should interact with a brand like Uber through conversation while providing a consistent and seamless experience.

“When I proposed this, there was some confusion,” Chris says. “Even though Uber is sort of on the bleeding edge of a lot of these things, the discipline of conversation design is still in its infancy.”

Conversational products are necessarily about starting an ongoing conversation  with users. Even some of the brightest struggle for the right icebreaker.

Breaking the ice

When comparing conventional product development to conversational product development, Chris pits agencies against Facebook.

“Agency work has a deadline,” Chris says. “The deadline is when the bot should be done. You launch it, you do a big event, and then you go out for drinks afterwards. By definition, a deadline is the death of the project.”

The problem with deadlines is more often than not, they signal the end of the product development process, when really the conversation is just beginning. Deadlines are the period at the end of the sentence, not an ongoing and emergent discourse with customers about what they need (even if they can’t describe it themselves).

Facebook is one of the best examples of this, because they are constantly rolling out features to different subsets of their audience in different parts of the world,” Chris says. “If you live in New Zealand, your experience of Facebook is going to be radically different than most everyone else’s. To me, they’re in constant dialogue with their users through qualitative and quantitative methods to explore what’s going to work.”

The deadline, as it exists today, puts too much stock in the success or failure of a simple feature. Chris thinks to effectively start conversations with users, products need to move past awkward and momentary missteps and keep moving forward.

“As code becomes cheaper, you can have a lot of projects going on at once, testing hypotheses that you can throw over the wall to users, and if your thesis proves invalid, you can chuck it and move on to other ideas,” he says. “Google rewrote itself every couple years. Its codebase was constantly being tweaked and pruned. In fact, we celebrated deleting code if we found ways to do things better or more efficiently.”

The ability to move fast—faster than ever—and make constant course adjustments is key. Frequently changing the subject, however, is not a balanced approached to building products. Instead, PMs and designers need to get used to anticipating, listening, and replying with minute product improvements. Users should rarely be surprised by something you change because you’re in the product flow with them, seeing things from their perspective and creating continuity in how you continually refine the expression of your product.

if you build ctt


Follow the conversation

Conversational products promise more freedom. And freedom’s just more fun. Right? Maybe not, especially given the challenges that freedom pose for people building the product.

“There’s that Canadian Olympian event,” Chris says. “Man, what’s that called? Oh right! Curling.

To Chris, curling is the perfect metaphor for this iterative product development process.

“Essentially, you’ve got this large silver puck (standing in for your product) and you want to get it to a specific endpoint down the lane,” he says. “Once your users start using the product, it takes on a mind of its own. So as the product designer, you’re constantly trying to stay one step ahead and making constant subtle adjustments by observing early adopters and analytics to roughly predict where people want to go — before they even know where they want to go. If you build the right flexibility into your product, your users will reflect their intentions in a way that you can’t anticipate.”

So much of product development, management, and design is trying to guess where users are going. But when your users give your product a mind of its own, how do you get it under control?

“If you’re not already a student of human behavior and deeply interested in how people speak, think, and express themselves, you better get ready to be!” he says. “Voice and messaging contexts allow for more self-defined expression. Because of that there’s a negotiable space, where the user expresses an intent and then the system has to evaluate how to handle it, ask for clarification, or fail.”

Intent is easier to express through words than a hamburger menu, but natural language is harder to interpret than a click. Over time, it gets easier by understanding context and observing user behavior.

“In 2D design contexts, you can watch a user interact with your app — scrolling, tapping buttons, typing,” Chris says. “In conversational product design, you have to look at the broader user context.”

Maybe, a user comes to your bot because a friend told her about it, but your bot has no context about her, like a conventional product might. Or a user wants something from your voice UX, but she can’t remember the options because there’s no feature map.

“You need to reiterate the jobs your conversational product can handle, and to fail gracefully and quickly, because the friction of failure grates on users differently than in apps,” Chris says.

Ultimately, curling is a team sport. Both the people behind the product and the ones using it want to move the stone forward. If users don’t pressure you to move the stone far enough, your competitors in other lanes will get ahead. If you overshoot your users, you’ll confuse them and leave them behind.

“For voice, you have to watch closely and help train the system to respond to unforeseen utterances, expressions, idioms, and other tricks of language,” Chris says. “You’ve got to watch how users interact with and explore the realm of your service, and listen closely to what they say. The good news is that when you fail, you can follow up in the same context and ask your user to re-articulate what they were trying to accomplish.”

The job of the conversational product is to engage users and to collaboratively establish shared context over time.

Closing the loop

The two big things you can do for your conversational product is to move fast and build in versatility. Both of these will not only create a more enjoyable user experience, but also inch your product closer and closer toward the desired outcomes your users are pursuing.

Ultimately, whether you’re building a conventional product or a conversational one, the question is how you get the richest user feedback as close to the context and moment when they were organically using your product,” Chris says. “And, once you’ve gathered that feedback, how will you turn it into actionable changes that make it into the product, improving it for everyone?

That means starting a conversation with your users — and keeping it going, by any means necessary. Charming as he is to speak with, Chris doesn’t have tactics ironed out for every one of these products. The future of the space is too big in his mind. He can’t tell you the endpoint of this conversation.

He’s just here to get it started.

whiter chris messina ctt


Get the latest from Mixpanel
This field is required.