The importance and effectiveness of data visualization in parsing big data and relaying it in an informative, efficient and attention-grabbing way is hardly lost on publishers. Many of them have started producing custom visualizations on newsworthy topics — but only a few do it at a scale and quality level that puts them in the category of “World Class.” One of them is Bloomberg. To get a rare inside look at how they work internally, how the visualizations they produce fit into the organization’s strategy and how they measure the performance of those visualizations, we interviewed team leaders Timothy L. O’Brien, publisher of Bloomberg View, Lisa Strausfeld, co-head of Bloomberg Visual Data, and interactive designer Jeremy Diamond.
Drew Skau, Visualization Architect at Visually: Bloomberg View has been creating some amazing visualizations. Can you tell us how these visualizations fit into Bloomberg’s overall strategy and goals?
There’s a broader strategy for Bloomberg and then there’s a related, but specific strategy for Bloomberg View and for Bloomberg Visual Data. When I came aboard, we wanted to make data visualizations a cornerstone of the visual and journalistic signature of the View.
I reached out to Lisa before I even started, and told her that we would like to do a partnership with the data visualization team around robust visualizations with a point of view and that were explanatory or analytic and almost featurey along with a daily flow of data attached to individual stories. That was about a year ago and Lisa and her team have partnered with us about four or five times for really big ones and then on routine graphics that we do every day.
I’ve seen a lot of those and they seem to fit within a template. Do you have a templatized technical implementation behind these?
The team is about a year and a half old and growing. We produce as much as we can that’s news-driven, we’re also working with the vast data resources that are available to us at Bloomberg that we can make available to the public and to consumers. We’re kind of exploring this area between data and editorial. We’ve had our most fruitful collaboration so far with Bloomberg View. This recent data view experiment was really about an immersive, opinion-based data experience.
When our team started, we decided to put together a design spec for all of the charts and graphs that you might see, which didn’t really exist outside of the core terminal graphics that customers for the terminal product used. For public facing dot-com or other sources needed to have some sort of unification or some way to put their own stamp on some view of a chart. With that in mind, we’ve created a lot of graphics within that realm, and Data View is a deep dive into one of those charts and kind of bringing it to life and just trying to have different type of graphing elements come alive and let you interact.
When you set the initial design guidelines, this had to fit the branding of Bloomberg. Did you pull in other designers for this, or is it something your group did on their own?
It was definitely a collaboration with the team, but all of this was done as part of Bloomberg Visual Data. And Bloomberg Visual Data has its own visual design guidelines, but we have also been very involved in the design and identity of Bloomberg View.
It has been a great collaboration from our perspective because the parts are really co-dependent. The visual statement is reliant on the underlying data, and the underlying data would be less vibrant and robust were it not for the great visual statement that accompanies it. There is a lot of action between our editorial and data teams and it’s an authentic multi-platform interdisciplinary collaboration. I think it’s unusual. There aren’t a lot of media organizations that have the resources the in-house talent or the sense of purpose to pull these interactives off at this level. From an editorial standpoint, having people as talented as Lisa and Jeremy’s team in house is an incredible competitive advantage that’s not easily replicated.
Seeing the work that your group produces, you are in your own league. There aren’t many people that can do the caliber work that you release. From a team-structure perspective, these things all come together into pieces very well. Do you have multiple teams that work together, or is there one team that has multidisciplinary people on it?
Bloomberg Visual Data is just recently 14 people. We produce news-driven infographics, tools (like the charting tool that we provide for Bloomberg News and Bloomberg View), and we create live, updating evergreen products, such as Bloomberg Billionaires, Best and Worst, State by State. We’ve kind of created two teams. One is a news-driven graphics team that consists of five graphic journalists, and then we have a product team that’s designers and developers and prototypers, and everyone kind of works together. And then we have a super-talented team with capabilities ranging from journalism to design to prototyping to development, and then there are those rare unicorns like Jeremy that can do it all. We’ve collided these two cultures of a news-driven team and a software-driven team and we think that’s key to our ability to innovate new forms of data editorial experiences
The product team is most reliant on collaborations with editorial. Bloomberg View mostly works with the product team because we rely on them for the editorial piece. We learn from it, we sketch with it, we try to generalize it into platforms, templates, kind of reusable experiences.
And for us on the editorial side, collaborating with the data visualization crew sharpens our own thinking about how we can use data in more interesting and compelling ways in crafting a piece of analysis or an argument. Another great thing about the opportunity to collaborate is that it’s allowing us to experiment with new forms around opinion, commentary and explanatory journalism.
I can give one example. The visual data team has an R&D team, a development team. We have a charting tool that’s being used by about 70 journalists in Bloomberg News and Bloomberg View. The Bloomberg View writers write some of the most engaging headers of the charts and then the subheaders are the title of what the chart actually is. We’re interested in ways to scale this experience, and ways to templatize. When we see how these charting tools are used, it helps us figure out how to encode that editorial voice into the tool.
Do you ever touch sports visualization?
Last year we developed a Major League Baseball franchise valuation graphic using data put together by the Billionaires team. We’re also currently coming out with one for the World Cup. It will be a kind of bracketing, and there’s a BSports team here and they do predictive data. For any given match between two countries, they run a model 1000 times to yield the most likely outcome based on their data. They already have the whole World Cup modeled out and we’re finding a way to present that that will also allow the user to pick their own bracket and share that with friends. And then once the event is live, allow that to be updated with the outcomes with each of the matches and compare to what was there before.
Does your charting tool have the malleability to customize for things like sports coverage where you might want to match team colors?
It’s kind of open in terms of the kind of palette you can use.
Everyone seems to have a charting tool these days, Bloomberg’s differentiating feature is that the charting tool can use Bloomberg data directly from the terminal.
There’s an editorial side to these that involves an ideation process, a discovery process with someone analyzing the data. But often times you have the idea before you have the data and so you find the data that exists and see if it supports that idea.
What’s interesting is the “How Americans Die” visualization that ended up getting at least 800,000 unique views and 8.8 million pageviews to date. So it was a wildly successful data and journalistic product. It got retweeted by Bill Gates and Atul Gawande. As I think stories often do, it evolved. We originally (in the wake of Philip Seymour Hoffman’s death) wanted to look at drug-related deaths in the United States and then wanted to look at various trades (heroin, cocaine) that supported that phenomenon. Then our columnist on that piece of work, Matt Klein, took the initiative to go beyond that more narrow thought and look at causes of death more broadly. And as you started to look through the various sets of data, we began talking about it and it became a much more robust and wide-ranging exploration of mortality generally. Once that was together we put the data into an Excel spreadsheet with very crude but simple graphics that gestured toward what we were trying to convey narratively. We gave that to our collaborators in data visualization and they really spun it into something grand. They came back with some of their own editorial thoughts and questions and worked side by side with us editorially as well, so Jeremy and Lisa have been really pivotal.
Do you aim to produce a certain quantity of these per month, or are they just as they come up?
We have targets. As we get our sea legs on these, I would like to be able to do a couple of these bigger ones a month, and then there’s a whole bunch of more modest but important visualizations that we want to be able to do every day. But we want to do enough of them that people expect Bloomberg View is a repository of visual data, sensitized analysis and deeply embroidered explanatory journalism.
Are there other things like deep linking that you use to prolong the life of these projects, so that they don’t just get that initial traffic hit, but they continue to be valuable over time?
We do get consistent traffic on Billionaires (which gets updated daily). Best and Worst is always getting new rankings. State-by-state is connected to a lot of public data sources like the Bureau of Labor statistics and Industry Leader Board is updating multiple times a day, too. Our challenge and our mission is to find ways to take these products that get updated automatically and connect them to news. That connection between those slices of those data products and the news stories is a big part of what we focus on. We’re constantly figuring out new and better ways to make that connection.