Interview with Bloomberg insider: How things work at Visual Data team

How can designers, developers and journalists collaborate to create stunning data journalism products? Do journalists need to code? And how does a typical workflow for Bloomberg data visualizations look like? Read this interview with Jeremy Scott Diamond, a data visualization specialist for the Bloomberg Visual Data team, and you will know the answers.

Please tell about your workflow: How do you come up with ideas, how do you develop them and which tools do you use for that?

The Visual Data team is sort of an anomaly among news company graphics teams. Bloomberg being both news and data centric, we’ve positioned ourselves to work on a range of projects: short-cycle graphics, reusable interactive templates, and long-cycle updating products that can be sliced into a variety of views.

One of Bloombergs long-cycle updating products. Source:

One of Bloombergs long-cycle updating products. Source:

My work tends to be in the reusable and long-cycle areas. Ideas for projects have come from team brainstorms, sighting under-exposed resources from within the company’s many groups, company-wide initiatives, and just finding good data to work with.

We develop the concepts through a cycle of data collection, analysis, and prototyping. The data comes from many different sources and requires different levels of care. Sometimes this can be simply handled in Excel, but often I use Python to help with cleaning and munging. Some preliminary work can be done using R to chart and get a better sense of the data’s make-up. Once we have a basic concept, we start prototyping the interactive with Javascript while wireframing and working through the UX/UI process. From there we keep going back and forth until it feels right.

Do you team up with Bloomberg reporters for specific stories?

Obviously the news graphics component of the team has a more traditional model, working hand-in-hand with reporters. Often my work entails working with data domain experts, as well as reporters. Most of my projects are within a realm in which Bloomberg already has analysts working with the data on a regular basis, so collaboration is essential. Sometimes the expert is the reporter.


How Americans die. Source:

In the case of “How American’s Die” and other instances of that presentation which we call a “dataview”, I worked directly with Matthew Klein, a former Bloomberg View editor. He’s the kind of reporter who is constantly basing stories on data, so it was a natural move into creating the more immersive interactive graphic from his work. He would research a variety of sources, finding the most interesting moments to highlight and comparisons to make. This would all be composed in Excel and brought into a “dataview”, where we’d go back and forth on the sequence and pacing.

Do you think designers and journalists need to learn how to code?

Need to? No. Coding is not essential for either of those tasks. Stories are most often not hidden away in a database and designers are supposed to envision the world they think is most applicable and composed. Being able to create the medium in which these practices are applied is not fundamental. However, having the skills to work with these mediums and some fluency with their languages can aid in certain forms of research and exploration. I think collegiate programs should offer a course in R or Javascript and those who are inclined will find their way.

Example of reusable interactive templates at Bloo,berg Visual Data. Source:

Example of reusable interactive templates at Bloo,berg Visual Data. Source:

Ok, let’s talk about this extremely important trend of visualizing available data and finding new insights. What do you think is the most important thing in data visualization right now?

If something can be summed up in 5 charts, it’s a success. I’m kidding. I think the most important trend is the data news story. The FiveThirtyEight , Vox, or Quartz churn of output which, to me, seems like it’s fueling a need for quick-hit information which may, or may not be, self-perpetuating. Maybe it’s what’s on my radar, but it feels pervasive and like it’s influencing content.

What is most challenging in creating journalistic visualizations?

Finding the best way to carry the narrative and deadlines.

What are your favorite types of visualization and why?

Anything that allows me to see something in a new way (which could be as simple as bar chart). Often, it’s the data (and the applied perspective) that makes the difference. Like the NYTimes ‘Baseball’s Borders’ piece. Maps can be boring, but when you have a data source that doesn’t fit in the usual format, you lose the conventional form and get a different perspective on a city and its team’s span of influence.



What are the most important principles of data visualization?

If you want rules, Tufte can fulfill that need. I think one of my favorite quotes and a good guideline to go by comes from Amanda Cox and goes something like “It shouldn’t feel like homework”.

How important is it to tell a story and to create a visual narrative? What’s more important – to explain things or to let people explore?

Tools for exploration are useful, but are typically for experts or those who already have some knowledge and interest. All news graphics need narrative and most tools can have better reach and usage if they lead the user in with some amount of perspective and discoverability. Exploring has a high barrier of entry without narrative, so in the best case there is a balance.

2 comments for “Interview with Bloomberg insider: How things work at Visual Data team

Leave a Reply

Your email address will not be published. Required fields are marked *