Make Users Feel Like they’re Part of the Development Team and Make Conversational Technology Great.
Fallbacks Equal Failure
Englishmen like myself, apologise first and often in every situation, because we believe it’s the right and proper thing to do. Unfortunately, the law of diminishing marginal utility tells us otherwise.
The more we apologize, the less it matters.
Judging by how often and in how many obsequious ways your average chatbot fawningly says some version of
‘Sorry, please try again’
you’d be excused for assuming almost all bot builders were English. They’re not, of course, they’re just hopelessly addicted to apologizing. And here’s why.
1. Users Are Never Wrong
It’s an article of faith in tech that the user only makes mistakes because the team can’t code for shit, or failed to make the process clear. Doesn’t matter if the syntax is sublime and the UI simple and sensible, it’s never, ever, ever the user’s fault….even if it is.
2. The Rise and non-demise of the Ten Minute Chatbot.
Structuring and writing dialogue for a great chatbot takes intelligence, focus, creativity and a whole lot of time. Just one problem…
“Build a great chatbot in just ten days”
is not a recipe for overnight success. Botland easily dispensed with that problem in short order with a simple one word edit.
A one-word edit dealt with that problem in short order.
“Build a great chatbot in just ten minutes”
The Ten Minute Chatbot is almost as good as the Ten Minute Novel “. The ‘need for speed’ as the number one selling point led to disaster. Dead ends, missed keywords, and AI/NLP that had a lot of learning to do, meant, well, ‘Sorry, please try again’, again and again. Creating a bot that delivers takes planning, perseverance, and time.
3. We Buy Tech’s BS About Its New Stuff
Facebook Messenger bots were going to ‘kill’ Apps when they exploded onto the scene in the heady days of Spring 2017. Suddenly, bazillions of these app-killing bots were everywhere. Naturally, users expected these bots to kick serious ass. They didn’t.
Bots are Dead. Long Live Conversational Apps.
Three years later most bots are still terrible. Many chatbots say nothing but
‘Sorry, I didn’t get that. Try again’.
So it’s no surprise that a few weeks ago, thousands of these unloved orphans had their bots kicked off the Messenger platform.
“Don’t bother trying again”, jeered disappointed users, sick of all those ‘sorry’s’ shoved in their faces by badly-built bots. They vowed never to touch a chatbot again. Unfortunately, the moment they walked the walk, was the moment bots began to talk the talk, as new creative tools and more powerful AI came on stream. Chatbots were out, conversational apps of real value in. Everyone was now a conversation designer. We had to get those users back.
New Title. New Thinking.
The Englishman in me politely suggested that perhaps if we conversation designers and creative writers in the digital conversation game apologized an awful lot more we could get those poor be-knighted users to try out our new conversation apps. But the American in me (I lived in LA for many years) said we should quit apologising to the users, and ask whether they were actually part of the dang problem.
Users part of the problem?
The idea sounded positively sacrilegious, and remained so until a pattern emerged in the conversation logs.
Incomplete, incoherent, mangled, misspelled and downright lazy user responses to concise and straightforward output were overwhelming even the best AI, and breaking decent bots.
The irony was as thick as a good leek soup. The “sorry I didn’t get that” fallbacks users saw as annoying signs of our failure, were just as often triggered by their input that was too poor for the AI and NLU to catch.
The Englishman would apologise no more. Instead, he would do the thoroughly American thing and…
Motivate The User to Deliver Better Inputs
And here’s how.
We give our new conversational apps all the power of today’s AI, plus a great persona, pithy conversation, exciting content, and gamified rewards to drive better user input and give users collective ‘ownership’ of the application.
Pride is a great marketing device.
Once users know their application’s success relies on their input, a sense of achievement takes over, and the improvements that the users had directly and collectively contributed would be something worth bragging about to their friends and colleagues too.
Plus you just can’t beat human.
AI’s awesome and all, but in a conversational app it’s fueled by user inputs. Improving them will almost certainly mean a crowd-funded version of the humanisation which is the driving force of the people-centricity of this consultancy’s philosophy.
The Power of Agency
Agency? What’s that. Agency is a psychological term for feeling as though you’re making something happen, or helping to make something happen. It’s a good feeling, an empowering feeling. And you want more of it. Let’s apply that to Conversational Apps, and in particular, combine two key elements of the chatbot. First, that it has to be nurtured and iterated over time to be successful. The second is that User loyalty is critical to its success. Now let’s add Agency to the way the user feels when they use it. Let’s term that agency, being “part of the team” that created the app. You could easily and legitimately make the case that the users as a whole are, in many ways, the most important part. Now let’s make it standard, that good feedback and better inputs are rewarded in some way (it could be as simple as “power user” points” or as much as cash (on a high-value ecommerce conversational app), and your AI will thank you. Plus you get that all important win-win, with User loyalty building and the bot improving — creating the most wonderful feedback of all — the positive feedback loop.
Unlocking the mystery. The golden metrics of conversation Applications.
With a motivated user base, conversational apps would finally be able to deliver the publicly viewable metrics almost completely absent in the “chatbot” era, very probably because none of them are worth bragging about.
These “Golden Metrics”, like stellar engagement, user growth, user retention, installs, and per session usage could all be within reach. If we hit their KPIs, client buy-in would be way easier. The centrality of motivating the users to see themselves as part of the team (heck, even leading the team) could be the technique that changes everything.
The Achilles Heel of “Non-Con” “Apps exposed.
It’s happened to me and it’s probably happened to you too. An auto-update of our favorite “Non-Con-App” (NCA) downloads and installs, and when we hate it, we tweet-rage that we weren’t consulted.
Con-Apps (CAs) with fired-up users will never have that problem, which could help them go mainstream…
…and maybe, just maybe, those, safe, smug, NCAs will finally have a real battle on their hands.
Originally published at https://www.linkedin.com.
