Feels Bot
Worked with a team of editors to generate Facebook Messenger ‘bot’ idea, ‘Feels.’ Created conversational flows, wrote copy, designed daily charts, social cards and design on final visualization. Sketched out initial information architecture for how a user would interact with bot. Tested internally and interacted with users through Slack group. Won a Society for News Design Award of Excellence for Experimental Design.
This is Feels.
He is the face of our Facebook Messenger experiment, called Feels. We had a lot of fun with this one. Here's the concept: Every day for the three weeks leading up to the election, we asked our readers on Messenger to respond via five different emoji ranging from '😀' to '😡'. We then collected the responses and showed them visually where they stood in relation to other readers every morning, along with some long-form responses if a reader elected to share. I worked with a team of social media editors, a developer and a product manager to refine the approach and create a balance of collecting information and providing that to users.
Setting up
We knew going in that this would be a true experiment. 'Bots' are still in a nascent stage. There aren't a lot of KPIs aside from creating a unique experience that we can learn from, and potentially use for articles or research for future bot/AI experiments.
A small team of social media producers, a product manager and I sat down and created a conversational flow using MindMeister to map all potential user input interactions, and set up 'walls' or boundaries to keep users in the daily flow. We wanted to avoid having the bot do 'too much' — i.e., providing custom alerts and options in addition to the 'Feels' flow — and because we left those out, our engagement numbers were higher than our previous bot attempts.
Daily process
Each morning, a social media editor and I would monitor the previous night's quotes via a Slack integration that allowed us to see user feedback, select a few to share, and I'd whip up a graph with our emotional data. Not 'true' AI – I still had to get up early every day for 3 weeks, which was the hardest part – but because of our input/output flow the editors and I put together, we ran out of technical scope to automate graph delivery.
Marketing & Promotion
Created a series of visuals, social images and marketing materials around promotion of the experience. Created custom share images, multiple Instagram & Snapchat stories and a final visualization (which you can see here) that wraps up the project and incorporates the data we collected over the 30-day period. Worked with our product manager to determine the best marketing channels and create custom imagery specific to each one.
The logo
Not a lot of bots have their own logo, but in conversations we had as a group it felt like this experience could benefit from having a sort of 'digital face', since it straddled the line between being 'bot-like' and curated with a human hand. Around that time, we had been sharing the robot emoji on slack a bunch per day. I wanted the representation of this bot, Feels, to have some sort of robotic presence so users would understand that at its core it had limitations, like a robot. But I also wanted it to feel human, imperfect, a little rough around the edges. The concept I came up with was 'the iOS robot emoji as envisioned by 80s NYC street artist Keith Haring'. I drew a couple concepts, and wasn't really in love with what I came up with, so I approached expert doodler and fellow WaPo product designer Taylor Schena and she nailed it. Thus, lived our little robot guy.