Diversifying Data Storytelling: An Interview with Silvio DaSilva

In the last couple of months, photo sharing behemoth Instagram made headlines in its plan to remove the “Likes” count, a metric central to the user’s experience of their platform. For years of Instagram’s use, this single number on each picture offers users a quantifiable reading on the popularity, success, or likeability of their picture or account. With the rise of Influencer marketing, the “Likes” number provides brands with practically the quantitative measure to gauge ROI on their sponsored posts. 

Why? “We want your followers to focus on what you share, not how many likes your posts get,” Instagram says in one of the screenshots depicting the new Instagram without a “like” count. “During this test, only the person who shares a post will see the total number of likes it gets.” There was also talk around the mental health implications that “like” count inflicted on users, as researchers suggested users’ self-perception too often was measured by number likes their posts got. 

While this update would be a huge shift for Instagram, it also illuminates fundamental questions around storytelling in data, and how visualizations designed to lead a user on a specific journey can have other implications as well. 

Reflecting on this, we spoke to Silvio DaSilva, the VP of data visualization at Moody’s Analytics, the credit and risk assessment financial giant. With a background in media data vizzes and a professorship at Parson’s School of Design, DaSilva has shaken up the dreary charts of Moody’s tradition and introduced more artful, creative systems thinking in approaching visualizing financial data.  

This interview has been abridged and edited for clarity. 

Keen.io: As in the Instagram case, although other examples including Uber or FitBit, many platforms out there give the user one or two metrics.  It’s shown us that a data visualist’s oversimplification of a story a data can quickly lead to unintended consequences. So what does this say about simplifying a platform and the data you’re feeding the customer?

DaSilva: We’ve come to this problem because of several things: 

First of all, too many data scientists are not sufficiently trained using principles of design thinking. That includes everything, the whole process from data science to data visualization to data design to cosmetic design. 

The problem is that design schools are not teaching this. There are very many schools that are jumping into how to design data vizzes, but they’re not forming professionals that understand the implications of design. Instead, they’re forming technicians. 

You have to be a creative. Understanding the human implications of what a single metric can have means also working with all sides of your brain–both for artistic purposes and for empathizing with the audience you’re trying to reach. In the case of Instagram, the team is only doing the technician’s job now. 

These principles were pioneered by Jack Burtin, a French data visualist who developed this theory in the early 60s, which was adopted by his student Edward Tafty. He was my teacher. Today, he’s 87 years old. 

Keen.io: How can these visualizations work to better speak to the audience? 

Much of what information design means is information graphics, data viz, explainers, etc. 

If the audience is a consumer audience and they’re getting your visualizations on mobile, you have to simplify that. The visualization should be more like an explainer than any kind of graph. For the most part apps do this now. The reason is very simple: small fingers on a small screen won’t be able to navigate a large amount of information on a graph. I cannot tell you how often I see this mistake being made in my career. 

So, the question boils down to: how sensitive can you be to your audience?

For example, I’m working on visualizing the financial impact of the European immigration wave to an audience of CEOs and investors who are very aesthetically conservative. I cannot mess with the blue color pallet or standard typeface because it’s part of Moody’s brand. But I can make the brand more innovative and more elegant. No whistles and bells, just simple and straight to the point and the information, because this audience wants these visualizations in order to draw financial conclusions in the fastest way possible. Once you have that insight, you work backward from there. That’s empathizing with the audience you’re presenting to. 

Keen.io: What are some examples of ways you’ve changed up how Moody’s traditionally visualized financial data? 

DaSilva: When I started working at Moody’s, they used a lot of pie charts. People like pie charts. 

I changed that. We no longer do any pie charts because they’re not a real representation of information. You cannot put into a pie more than six data points, because it gets confusing. I’d sooner do bar graphs with lines and colors whose visual sum can depict a trend. 

By eliminating pie charts, we also developed a program using d3 from excel. I get d3 data that fits the size of excel enabling creation of better-looking charts. Without having to detach anything, the info feeds straight into the CMS. 

The other thing is about the boring bar graph – sometimes, I want that bar graphic in my reports. I want to see the graphs in context with the story. Using a regular table doesn’t scare me either, because sometimes it simply tells the story the best. Sometimes what the analysts or editor envisions is not always going to look better.

The same thing applies to Instagram or any other consumer app. Being unafraid to change up how they’re depicting their data can mean leading the user on a more nuanced journey that’s best for them, and will keep them engaged with the platform. 

Key to all data visualization, says DaSilva, is carefully curating the story your platform’s telling with data.

To learn more about how to begin visualizing data meaningfully, speak to a data visualization expert at Keen.io by booking a demo here