Why type when you can tap?
Are chatbots the next wearables? I’m trying to understand the potential of chatbots and how people engage with them, and more importantly, why? Why is it so compelling to go back to clunky text input when we’ve reached peak swipe?
Use Chat bots to get 80% conversion rate on email sign ups.
Why type when you can tap?
I’m trying to understand the potential of chatbots and how people engage with them, and more importantly, why? What’s the use case? Why is it so compelling to go back to clunky text input when we’ve reached peak swipe? We’ve spent the last 10 years perfecting tapping an icon to get instant access to solve the problem at hand. Why go back? Or is it back? Or is a ‘millennial thing’ — the place we put all shiny new things we’re trying to figure out?
Building the Bot
I like to learn by doing. Not always the most efficient way, but so what. Sometimes you gotta get your hands dirty.
Assuming Facebook and others know more than I do about why they’re doing this, and clearly they do, I decided to have a go at building a bot in messenger. (Experiment 1)
I’m a lightweight hacker at best (by that I mean I last wrote proper code 15 years ago), so I picked ChatFuel as my botbuilder. No good reason, I found it first and it works.
What I wanted was
- a drag and drop interface,
- external hooks into Zapier so I could see if people would leave their email addresses and at what rate (Experiment 2),
- and to give away a T-shirt I designed for myself. (Experiment 3)
Enter Elon Musk and Exploding Rockets
I picked a topic close to my heart. Elon Musk’s incredible life journey that is currently launching more rockets than most governments. While building Tesla.
The bot is a storyline starting with Elon’s decision to start Zip2 and engaging the user to make a few decisions along the way to finally launch a rocket without exploding.
What is a bot?
Basically it’s a few lines of text, some free gifs from Giphy, and a Zapier hook into Mailchimp for the subscribers and T-shirt entries. And some clever canned ‘AI’ responses provided by ChatFuel.
This connects to a dedicated Facebook Page to host the Message bot. Setting up a dedicated bot page for Facebook was simple and Chatfuel publishes directly to the page.
It took me a weekend to get it done and ‘launched’. I then put it on ProductHunt and Botlist and waited. Within a few minutes people started playing. This was a few days ago so the stats are still coming in.
You can try the Ride the Dragon with Elon Musk bot here.
Observations
- Most people leave their email address. Around 80%. That’s almost impossible to do via an app or a web based solution. I suspect people trust the messaging space but I’m not 100% sure why this is.
- People text to the bot as if it’s human. They’ll say ‘please’ and ‘thanks’, etc during interactions.
- Very few abuse the bot with profanity, etc.
- This particular use case is trivial, clearly, so creating a ‘sticky’ bot would be tricky. These are all drive-by users. I suspect creating a constant stream of fun bots around a subject could compound the engagement over time. Or a decent use case that’s not a game.
Are there useful bots solving problems?
Here is the MAKEUSEOF 10 top bots list. Have a look.
The challenge I have is all of them are either subsets of apps that already do a simpler, better job with less taps, or they’re trying to add humour to things like the weather — Poncho for example. This is funny for a few days and then just annoying. Who wants another stream of garbage to look at? Facebook already does a great job at this.
Are bots the next wearables?
After spending 18 months unsuccessfully building a wearable app and watching the technology search desperately for a problem t0 solve, and wearing my $1000 notification system that takes 2 seconds to wake up to show me the time (otherwise known as the Apple Watch), I’m skeptical.
Are messenger bots or chatbots going to find the problem to solve, and solve it in a way that is 10x better than we currently solve it, or is this simply the evolution of smart natural language processing to something more useful, like taking the call load off your call center staff at unpredictable peak times? I suspect the latter.
I’m happy to be completely wrong here, so I’d love to hear if you’ve found a bot that works well and has replaced another behaviour to get a better result.
☞ Please tap or click “♥︎” to recommend this article to others.