“Companies find it hard to get hands on their own data”
Marketing departments demand fancy visualizations and interactive graphics, but they don’t realize they first need the data. An interview with Katy Harris and Mark Schifferli from the design studio Fathom.
May 18, 2012
Fathom is a Boston-based design and software consultancy that has produced data visualizations for prestigious organizations such as the Gates Foundation and General Electric. The company was launched in 2010 by Ben Fry, a world-renowned expert in data visualization whose work has been shown at the Museum of Modern Art in New York and featured in science fiction movies such as Minority Report and Hulk. While still a graduate student at the MIT Media Lab, Fry co-developed Processing, a programming language specifically built for electronic arts and visual design.
In March 2012, I met two members of Fathom’s team: Katy Harris, a designer trained at the Rhode Island School of Design, and Mark Schifferli, a developer who is passionate about working on large and complicated datasets. Our conversation took place on an abnormally hot day, in a café near their office on the edge of Boston’s historic red-brick Beacon Hill district.
You’re specialized in data visualization. Is the demand from companies growing?
Katy Harris: I think the demand is growing, but there’s still a data bottleneck within large companies. People are excited about visualization, but find it hard to get their hands on their own data.
You have to push them to obtain it?
Mark Schifferli: In fact the push is often internal to the companies. Large corporations are often structured in such a way that data isn’t always available for sharing between departments, not to mention to consultancies like ours. Obviously, privacy concerns limit the data that is available to us. Beyond that, often a client’s first instinct is to share only the data that has been pre-aggregated, but if that’s heavy handed enough we run into complications where the data no longer tells a distinctive story.
There aren’t many studios like yours who specialize in producing data visualizations for the private sector.
Katy Harris: It’s actually a challenge sometimes. The field is so new that there aren’t many real precedents to show prospective clients. We have to explain to them what is possible and what isn’t – and we don’t always find the right words.
Mark Schifferli: We often end up creating prototypes for our clients to demonstrate just how effective visualizations can be. To that end, we recently made an example based on the evolution of the Fortune 500 over the last 55 years. The beautiful thing about that piece is that it’s extremely dense in terms of the information it contains. It takes only half a computer screen online but more than 50 pages to print the data.
We also like to emphasize how interactive visualizations can communicate more about the data than static infographics. Sometimes a single image is all you need, but interactive applications allow multiple views of the data, which can reveal more about the underlying information.
What’s your philosophy? Should the reader always be able to clearly understand the data behind your visualizations?
Mark Schifferli: That’s the goal. We want to communicate. We want to be able to present the forest as well as the trees; the take-home message as well as all the detail.
You’ve created a dozen or so impressive data visualizations for GE. I found your recent ones about the gas turbines and the medical scans more difficult to understand.
Mark Schifferli: Well, the web version is actually just a preview; in the real world, it is a massive touch screen installation. Right now it sits in GE’s lobby but it’s going to travel around. Unlike online movies, these visualizations are interactive: people can zoom into specific areas in the data, stop the animation, move it, etc. It’s a whole different story from what’s on the web.
Katy Harris: Audience and context really play a role here, too. Those two in particular were designed for interaction in a physical space, for an audience that is generally walking past without a ton of time. It needed to be compelling graphically from very far away, and to reward you for moving closer by revealing more detail.
How will GE use it?
Mark Schifferli: They have access to a wealth of data, and are interested in using visualization to demonstrate their expertise and further the conversation around issues like energy production and medical imaging. GE has a strong appreciation for design and a tradition of communicating with the general public. We’ve produced a lot of visualizations for them based on publicly available data. Often, websites distribute data in a format that is very disorganized and hard to navigate. Our job is to make it easy to consume. We hope that the average person will be able to appreciate and understand it.
Is GE your main client?
Katy Harris: GE has been a great partner. When we started, we produced a few smaller interactive pieces, but have shifted into visualizing more complex data sets. Most of our current client work tends to be either for tablets or touch screens – that’s what really excites us right now.
Mark Schifferli: The lifespan of something on the web can be very short, since browsers and web technologies change so rapidly…
What’s your design process?
Katy Harris: We don’t start the design process without the data. Then we work iteratively together. After some group discussion, Mark might create an initial sketch of the data in Processing and check it into our repository. I generate vectors from that sketch and refine the visual design, which he adapts back into the shared code. From that point on we both iterate primarily in code, and we don’t stop changing things until the very end. It’s pretty fluid.
Do you try to find interesting patterns in the data?
Mark Schifferli: It depends on the client. Some clients have a deep understanding of their data and want a tool to really explore and get at every facet of it. Sometimes the client wants us to find the story in the data, and then we go exploring. And sometimes we run into cases where the client has a idea of want they want to show, but the data doesn’t support it.
Katy Harris: We also make our clients work! There is a lot of back and forth, asking questions about the way data is collected, or pushing a client to include more information or augment it with complementary data. We work together to arrive at the most compelling data we can.