August 30, 2016

What's Wrong With Bots & Conversational UX — And How To Fix It

Kurt Leafstrand

Kurt Leafstrand

Kurt Leafstrand
Kurt Leafstrand

VP Products, Clari

B2B Banner1380X600

If you’ve ordered a pair of pants via bot on Facebook Messenger (yes, all three of you), or “read” the news for more than a few minutes on the Quartz app (and don’t get me wrong, Quartz — your content is unparalleled), you know virtual conversations can be even more tedious than your seatmate on the redeye who just can’t stop talking.

Seldom has the gap between hype and reality for a new technology been so great, even by Silicon Valley standards.

Early approaches have failed because they have not fully understood the nature (and nuance) of this new medium. They assume people like conversation for its own sake, and in the process, take formerly simple tasks and make them several times more complicated. Just try to use a bot to buy a pair of pants.


Before:



After:


Hats off to early bot pioneers like Spring who are exploring this rapidly-evolving paradigm, but this is not your parent’s Clippy. New thinking is needed to get conversational interfaces right. So how can you build a product that fully exploits its strengths?

Conversations Rise Above the Noise

The marriage of conversations with intelligent notifications is the most promising path yet to managing information overload. It’s a format ideally suited to curation: email is too noisy, apps are too passive.

Strong conversational products think deeply about curation. They invest heavily in bots that don’t just walk you through a set of questions, but integrate with a intelligent cloud services to sift through everything important to you and surface just the right nuggets at the right time to be valuable and actionable. Rather than being a time sink (love wading through your email inbox after vacation, anyone?), good bots accelerate “getting things done.” And conversations let them do through a medium that’s far more intuitive to users than previous attempts (RIP Google Now).

Conversations Enable “Machine-Assisted Actions”

Before intelligent messaging, accessing information and conversing about it were two different things. Sure, Facebook and derivative interfaces supported liking and simple dialogs around status updates, but the current revolution marries conversations with powerful machine learning and data science. The result is human conversations and decisions prompted by new insights.

Take a critical business process like performance evaluation. Historically, this has been a tedious process where managers must remember a year’s worth of employee activity in an attempt to generate a constructive set of feedback and recommendations for improvement. Even companies who sell HR software are ditching them. With a messaging-based interface powered by machine intelligence, a simple mobile app can automatically prompt managers and/or coworkers to provide feedback on an ongoing basis and intelligently route that feedback to employees to drive immediate action and course-correction (see companies like 15Five). Messaging + AI promises to both simplify and improve countless business processes in this way.

Context is King

The best conversational interfaces dynamically create and destroy messaging “contexts” to combat information overload. Any Snapchat user understands the incredible power of ephemerality (hint: it doesn’t really have anything to do with sexting). Again, consider your inbox: it will grow forever if you don’t maintain it. The beauty of ephemeral conversations is the user expectation that they are actively throttled and managed to surface only what’s relevant. Didn’t look at your friend’s story within 24 hours? It’s gone, because it’s old news (especially if you’re a teenager). 

Enterprises can learn a ton from this consumer trend. As companies gain the ability to control and manage what gets their employees’ attention at any given time based on the company’s strategic priorities, the result will be transformative. Combined with notifications and real-time communication, this intelligent curation will dramatically improve worker productivity not just by making things easier, but by helping people maintain focus in a world of ever-increasing distractions. 

Look at what companies like Crew are doing to change the way highly “tech-underserved” segments of the economy work by leveraging this concept. On-the-go workers like restaurant employees can easily manage their shift schedules and get critical, timely information without having to get email addresses and constantly manage their inboxes. Uber, while not a conversational interface, can largely thank the power of context and ephemerality for its success, since those concepts are what make it frictionless for drivers and riders to see in real-time the availability of transportation in a given geographic area.

Finally, great conversational products understand that conversations are not a panacea. There is a complexity “tipping point” beyond which a messaging-based interface makes life more difficult — essentially, when a long conversation is required to accomplish the task. This can be due to the specificity of what you’re looking for/trying to do (like buying khakis) or the iterative nature of the process (like booking a flight). If it’s something that is frustrating doing over the phone with a customer service rep, it’s probably going to be frustrating doing it with a bot. 

Still, it’s early days. With such great promise and so much interest and brainpower focused on the challenge, the way humans interact with machines will be profoundly impacted by this new, digital art of conversation in ways that no one would have anticipated just a couple of years ago.


Stay up to date with all our latest resources, blog posts, and news.