What happens when you come back to Twitter after a day, month, or year?
In mid 2012, I left IDEO to join Twitter as the second designer in the recently opened New York location, situated on 43rd Street and Madison Ave in Manhattan. My mandate upon joining was to work with (my friend and wonderfully talented designer) Yaron Schoen on building out the future of Twitter Discover – a feature which has disappeared, only to be recently blended into other aspects of Twitter products.
Twitter for me was an incredibly interesting prospect. My good friend Andrei Herasimchuk was design director of the growing SF-based team, and Doug Bowman, a designer I had admired for ages, was creative director. Never mind the fact that I was (and still am) a giant fan of the platform. I signed up for Twitter in 2008, and I’ve stayed continually active ever since.
I joined the company with the intention to help ship incredible products to over 260 million people all across the world. Most of the work that I was doing until that point was unsharable due to NDA, and here was an opportunity to publish with frequency. Ironically, much of the work that I accomplished at Twitter didn’t ship directly. I like to think, however, that it had influence.
In late 2012, my good friend Mike Davidson joined Twitter as VP of design. He reached out to me with an interesting question: how can Twitter help people get caught up with what they care about most after they've been away for some time? Twitter didn't offer this feature, and he was hopeful the New York team could help.
Twitter thrives on data. When I was at the company, the weekly Friday global all-hands meeting contained a reoccurring segment entitled “We Measure Things,” where we’d discuss product metrics and how they subsequently impacted the business. Unsurprisingly, Twitter had data that could answer pretty much any question about itself. We were quick to investigate just how much stuff people miss on their timeline. Let’s just say the numbers were higher than expected – people miss a tremendous amount of content every day.
This understanding served as the core of the work that would subsequently come next. If people miss a lot of what they might care most about, then Twitter should be able to find a way to serve those meaningful tweets up to them in a better way. But how do you design the system so that it doesn’t affect how Twitter is fundamentally understood: a real-time, reverse chronological firehose. And besides, what do people “care most about” anyway?
By attempting to understand the latent needs, wants, and desires of the people who use our products, we could identify opportunity areas for design.
Early analysis led us to a hunch: perhaps not all tweets were equal. Some accounts and tweets are more important than others. If we could understand certain aspects of people’s behavior on Twitter, perhaps we could separate the meaningful signal from the noise. By attempting to understand the latent needs, wants, and desires of the people who use our products, we could identify opportunity areas for design.
We would spearhead the work in New York, working closely with select people in San Francisco to keep communications flowing. I was leading the project from the east coast, working with some of the most talented people I’ve ever had the opportunity to work with in my career. The team in New York was small, but robust. In house, we had both Erin Nolan and Dave Desandro. In San Francisco, my friend Josh Brewer kept us sane and connected to the mothership.
The project started with a timeline. I proposed a 3 month project cycle that involved just enough research with subject matter experts, appropriate time for download and synthesis, the ability to gain early real-world feedback and proper time for prototyping and detailed design. Thankfully we received a go-for-launch. We named the project “Recap.”
We kicked off the Recap project with a group brainstorm on the topic of “catching up” in the broadest sense possible. We attempted to identify people, processes, software, services, situations, and jobs where the major use patterns involved getting caught up. We decided that the best way to begin understanding the fabric of catching up was to speak with people in the field of information overload, and with those who understood the nature of “sticky” information.
Our team recruited a small number of industry experts who were willing to speak with us. We planned on-site interviews where we extensively recorded and collected detailed information. We completed the entire research phase in a single week – including capture and design synthesis.
The latter half of the week proved particularly interesting. Most of the team had never gone through a synthesis phase until that point – and the results were terrific. I had the ability to both lead the process and be taken aback by different approaches at the same time.
The patterns that emerged via synthesis proved to be an amazing launching point for the next phase of the project. We immediately got to work roughly sketching out as many ideas as possible, instead of allowing ourselves to be seduced by a single “right” solution. What you see above are samples of the work that came from that process.
The ideas that we sketched quickly began to coalesce into a few clear design directions. We mocked up these directions using Balsamiq, and made rough but considered click-through prototypes.
We baked into our schedule time to put some of this early work in front of actual Twitter users in order to test our ideas and directions. We ran into some challenges, however, putting work in front of non-employees because of company confidentiality policies. Due to the fast nature of project we didn’t have the time nor the means to go through a lengthy recruiting process, which led us to believe we might need to abandon this step in the process.
We immediately got to work roughly sketching out as many ideas as possible, instead of allowing ourselves to be seduced by a single “right” solution.
Talking through our problem with one of my coworkers in engineering yielded an unexpected solution. I learned of a underutilized system where temp workers who already possess Twitter NDAs visit the office and review new engineering directions. Even better, this system was already part of the budget. Because of this, we managed to bring in a significant number of temp workers from around the greater New York City area to review our mid-cycle work. We shared directions, and carefully took note of successes and failures in order to further converge our designs.
Unfortunately, this is where I have to stop the story. I can’t show any more work, and I can’t really explain the system that we designed which would eventually become the Recap prototype. Recap in-full never truly shipped. There were a few reasons why – one of which (mirroring my experiences at Airbnb) was due to the fact that Twitter runs and extraordinary number internal experiments. Far from everything that is designed and built finds its way into the final product.
We also faced unexpected internal competition which further complicated matters. Additionally, the structure of the Product organization (which at the time Design reported into) was experiencing substantial shifts in structure and leadership. I can tell you, however, approximately what happened next.
We built an engineering prototype that we believed addressed our core question head-on, and provided substantial benefit to Twitter users returning to the platform.
Me and the rest of the Recap team created a framework based on the most resonant qualities of the directions we had been pursuing. This framework was research-led and user-insight driven, and held true to design principles that emerged from our work. We built an engineering prototype that we believed addressed our core question head-on, and provided substantial benefit to Twitter users returning to the platform. Because of changes in company direction, we as a team unfortunately needed to change course ourselves, and we returned to working on core (versus experimental) Twitter products.
A silver lining emerged a few years later, when the company shipped a streamlined version of Recap for mobile. The New York Times called attention to the feature, and noted that “Twitter is delivering on one of the promises it made to investors in November, when it previewed the “while you were away” idea along with other user improvements.” I’m happy to see that parts of what we designed will live on in the product, and I’m bullish that Twitter will continue to improve and invest in Recap over time.
What assurances do people need in order to have a great VR experience?
Applying the Airbnb model in different and unique ways.
What happens when you come back to Twitter after a day, month, or year?
IDEO helped design a more transparent genomic sequencer.
Creating and productizing a knowledge-sharing system at IDEO
IDEO is one of IDEO's most challenging clients.
Insights about people and markets help advertisers create powerful messaging that resonates.
How do you approach designing a social network for social workers?