Earlier this year, we partnered with a startup in San Francisco called Prismatic. We were asked to help them redesign their product. We got up to speed and worked alongside their designers and engineers refining some of the Web application before moving onto iOS. Today, some of that work came to fruition in the form of their newly launched iOS app. We plan on writing a full case study on our site in the new year, but I wanted to share some of the background, challenges and work we did in the meantime.
Some background
Prismatic started out back in 2011 as a really smart personalized newsreader for the Web and iOS. I became a pretty active user right from the beginning. It just seemed to implicitly know what I was interested in. Aside from being a good destination to kill time or dig deep on an interest, it was a good place for me to find stuff I could share with my personal network. Like many new products, it wasn’t without its faults. The article parsing, while clean and awesome when it worked, would sometimes fail. The typography was pretty rough in the early days, and the UI was trying to do too much on the Web—all at the expense of the content. Despite its shortcomings, I still found myself using it often.
Summer of 2012
I was in San Francisco working on another project when a coworker suggested we meet up for a drink with one of the founders of Prismatic, Bradford Cross. A couple beers later, he asked me what I thought of his product. I gave him a pretty honest personal assessment. I was impressed by how it would show me things I was interested in, like architecture, cars or food, even though 95% of what I follow and talk about is related to design—it felt a bit magical. I went on to tell him that I disliked the Web experience. That the interface was trying to do too much at the expense of the content. I believe at one point I may have said, “The headlines are taking a shit-kicking”. The iPhone app was better, but far from perfect too. I kind of just put up with everything about the apps because I was finding great content with very little friction. I’d later learn that Bradford is the type of guy who appreciates this type of honesty.
After talking about some of those details, he enlightened me on a much more ambitious goal they had beyond that of any typeface or UI system: They wanted to become the home for my full spectrum of interests. That’s a pretty ambitious goal, that they’d only begun to develop. Done properly, it struck me to be on the scale of that of Facebook or Google. What impressed me at the time wasn’t that they were thinking that big, but that they seemed to have the smarts to execute on it. Bradford expanded on what he meant by saying:
Current social networks box you in. If you’re into tech, but also into cooking and hiking, it’s hard to follow enough cooking and hiking people, and your followers are probably expecting tech from you, so there’s social pressure to avoid sharing about cooking and hiking. We want Prismatic to be way more scalable and elegantly solve for that. We want it to be the social network for interests.
You can follow all your non-tech hobbies and passions you don’t talk about on Facebook and Twitter, without needing to manage crazy groups or project a single persona. We can do it by ranking everything based on you and your network, so you always see the most interesting things at the top.
It immediately made sense to me. I’m into design. As a result, who and what I follow on Twitter, and who tends to follow me back, are also interested in design. While design might be my main interest, it’s not the only thing that I’m interested in. Most of the time, I end up not sharing content like how ‘Porsche’s 918 gets better gas mileage than a Prius’ or ‘the ugliest house in America’ because it would be considered off topic for the majority of my network. In most social networks, the primary interest of who you follow is directly related to the type of content you’ll get from them. I liked that Prismatic was thinking very differently about this.
Bradford and his team recognized this wasn’t something they’d fully solve in an iteration or two. This was incredibly complex work—both to think through, and to execute. It was, and is going to continue to take, a lot of smart people, time, explorations, research and iteration. We wanted to help.
Product design is hard…
Identifying and extracting content from a gazillion different places is hard. Their job here is to crawl and index all the interesting content on the web—social, mobile, commerce, media etc. In fact, at the time of this post, their system looks at 5 million new stories every day that contain over 10,000 interests. Hard machine learning then needs to identify and extract content and interest classifications. They need to understand what each piece of content is about and what genre it belongs to. They also look at personalized ranking based on the historical interactions of you and your network. When you layer in things like scaling and performance, you beginning to understand the complexities they’re dealing with.
…really hard
In addition to the technical challenges, these kinds of systems are awfully difficult to design for too. Mikael, Nate, Francisco and Gina, a few of Prismatic’s designers, know this all too well. One of the biggest challenges here is that the ‘stuff’ you’re dealing with approaches infinity in terms of variability. Think about it—You need to design a system that works for every piece of content ever written in the past, but also has the flexibility to work with content that has yet to be written. At the same time you need to balance things like interactions, commenting, what’s inside the system, what’s outside the system. In the truest sense of the word, it’s awesome. With any hope, you at least gain a new respect for the people behind those massive systems so many designers love to hate on.
Content variables are plentiful, here are just a few:
- Media type articles, videos, audio
- Topics it belongs to
- Associated media
- Descriptions or summary abstracts
- The body of content
- Content source
- Your connection to the content (people, interests etc.)
- Actions taken (likes, comments, faves etc.)
Other factors like the length of headlines and the contrast of images and other media can play havoc with making everything work harmoniously. You constantly need to ask yourself: Does the content stand out enough? Are the right actions available at the right time? Are they obvious? Are the distinct pieces of content tied closely enough together to look like a continuous feed, but separate enough to look like distinct? Can I scan it? Can I read it? Is the design ownable? Is it too generic? Does the design detract from the content? Does the design get in the way of the experience , or enhance it?
As a result we worked with their team of talented designers and engineers for months on this. As I was writing this, I began looking back at the archives of files—something I rarely do until we create deep case studies. It’s mind boggling how many options, directions and iterations we explored.
The big takeaway
In the end, a successful project is never done. It is never perfect. If you aren’t learning from it, then you’ve given up. It’s a constant process of assessing the landscape, making hard choices and accepting trade-offs. It takes passion, dedication, design, engineering, planning, talent, time, iteration, research, occasional moments of blind faith and a dozen other things I’ve forgotten to mention. If you’re lucky enough to have all those things, you need only rely on perseverance, though a dash of patience and funding doesn’t hurt. Today is an enormous step forward for them and their product. It represents these things and whole lot more to me personally.
What we love about working with people like Prismatic, is that they inherently understand all of these things. The product is, in my obviously biased opinion, a great example of very progressive iteration. It represents good change and a big step in the right direction.
On behalf of everyone here, I want to wish the fine people at Prismatic all the best with this chapter of their product and the future ones they are about to write. I thank them for trusting us in part with their livelihood. We had an amazing time working with you on it, and we can’t wait to see what the future holds.
Go grab the iOS app. It’s only the beginning, and it’s going to be awesome. You can check out Prismatic on the Web too.