Webby for Good is a collaborative program formed by The Webby Awards and WP Engine to showcase Webby-recognized projects built to change the world.
On launch night, our chatbot got so much traction that our discount serving server crashed. - TRUE Team
What’s the elevator pitch for your project?
The Sober Self Bot is the world’s first drunk driving campaign that sent night-outers messages from their sober selves.
“Nothing good ever happens after 3 am.’’ Mike Clement, Auckland Police
We created a chatbot through Messenger where people could plan their night out by setting a curfew and send their future (drunk) selves a message that would remind them to stick to it.
When our ""night outer"" chose to leave on time, a simple click within Messenger would get them home by taking them straight to their Uber account.
The nice bit - the Sober Self Bot would incentivise leaving with a 25% Uber discount.
The clever bit - the discount only lasted 15 mins. Encouraging them to stick to their word.
What was the impetus for this project? What real-world challenge were you trying to solve?
It’s not the drinking. It’s how we’re drinking.
New Zealand has one of the worst binge drinking cultures of anywhere in the world.
The NZ Drug Foundation says alcohol causes more harm in New Zealand than all illegal drugs combined.
The brief came in and the task was HUGE.
How do you change the behaviour of stubborn young kiwis, getting them to drink responsibly and head home at a reasonable hour, safely? Traditional shock and awe tactics through big TV ads wouldn’t work. A new more relevant channel needed to be identified and a partnership with ride sharing app Uber formed part of the client brief.
Our answer was mobile + a clear “what’s in it for me?”
No one knows your limits like you do.
We uncovered that while young people want to drink less, they didn’t want traditional "help" in doing so. People reported that they were more likely to use an app or online platform to seek help to reduce their drinking than through face-to-face interactions.
It was clear.
A predictable, parental drink-driving message wouldn’t cut it.
Once you settled on your idea, what was your first step in moving it forward?
Prototype. Prototype. Prototype.
We asked ourselves, what would make us engage?
We did this through messaging each other on Facebook Messenger to see how the conversation flowed. This was really important because while something sounded good on a word document, as soon as it was in the chat context, it felt clunky.
Once we figured out the conversation flow, we then worked on giving the bot its own personality – someone you would want to have a chat with.
Was there a moment during the project where you ran into a hurdle; or faced a problem you didn’t know how to solve? Take us to this moment, what happened and what did you do next?
On launch night, our chatbot got so much traction that our discount serving server crashed. While the developers were spinning up more server instances, we had a team of people performing a manual override. We had one spreadsheet of users and their go home times, and another one with discount codes of varying values that expired at different times. The great thing was that there was no downtime for users interacting with the bot, only a slightly longer response time while a person reviewed the spreadsheet.
Did the real-world impact meet your expectations? Can you share an example?
In the morning, we asked people how their night was. We received some great messages from people who said the bot helped them get home safely.
“We’re home! Thank you so much”
“Going great thanks” (in reply to our good morning message)
“Ok thanks. I had a great night btw”
What technologies/media did you use to develop this project (AI, Social Media, WordPress, etc.)?
Our bot was built on top of Google’s Dialogflow and then connected to Facebook Messenger. We then connected this up to Google Analytics so we could measure all of the activity happening on the bot. To drive awareness, we had ads on Facebook and Instagram, that changed out depending on time and day of the week.
Was the tech/medium you chose crucial to conveying your message? If so, why?
We initially thought this idea could be executed in an alarm clock like app. We then spent some time thinking about what the desired outcome would be and found Messenger was the perfect place to have this experience. This was because it was easier to link people into the experience from ads, it did not involve downloading an app, or require people to sign in.
How did this project defy your expectations?
We started to change New Zealand’s drinking culture. 60% of people used the bot more than once, and it became part of their habit when going out. We actually created a behaviour that continued after our paid campaign finished, which is a great result. It provides validation that gamification and behavioural economics actually do change behaviour, even for things that are part of our culture.
How will you use technology in future work to create inspiring, cutting-edge projects that also make a difference in people’s lives?
We believe technology has great potential to change behaviours for good as well as bad. We’re constantly prototyping and exploring ideas to continue the mission of Cheers! and hope to get more into market for them. There is always the barrier of cost, but this is coming down quickly as SASS tools allow us to get started quickly and not re-invent the wheel.