Visual analytics 101: When to use numbers vs. visualizations
Nelson Davis |July 21, 2014
This is the first post in our data visualization design series, which covers the essential building blocks to designing meaningful data visualizations that engage, inform, and empower your audience.
In its traditional form, data is nothing more than letters and numbers in rows and columns. Bringing that data to life—where it can be sliced and diced, interacted with, and ultimately consumed and understood—has become a science in and of itself. And while our data may begin with raw numbers, today we’ll ask when it should end with numbers and when it’s better to use a data visualization instead.
The fundamentals of visual analytics
The following is a great mind trick to grasp the basic fundamentals of visual analytics that I learned at a presentation. Showing the image below, the presenter asked the room, “How many nines do you see?"
The group squinted at the image for a bit, and someone in the room shouted, “6… no 7!” The presenter smiled and moved to the next slide, asking, “How about now?”
Ten. There are ten nines. Well of course there are—now they’re easy to see. This is visual analytics in its simplest form—using color for “needle in the haystack” analysis. Next comes a more real-world example of visual analytics: a crosstab of numbers, showing sales and profit for various parts of a company. If you’ve ever heard of Excel, you’ve seen a crosstab just like this one:
Next, someone asks, “Who has the best profit and who has the worst sales?” More squinting. Thirteen rows of data have brought the group to the edge of what’s possible to understand in 20 seconds. To the group’s relief, the crosstab is replaced by the grouped bar chart below:
Bar charts just so happen to be one of the best visualizations for comparing data. Knowing the order of east region sales by Product Type, highest to lowest (Coffee, Espresso, Herbal Tea, and Tea), is much more important than knowing the actual numbers for any one metric (e.g., sales in the west region for herbal tea are $72,285). These examples get to the heart of this discussion: the point isn't just to have the data; the point is to understand the data, create insights, and make better decisions faster. Let’s discuss how to create these data-driven insights for a user through proper uses of numbers and careful selections of visualizations.
Using numbers for precision rather than context
It is often said that the three rules of real estate are “location, location, location.” If that’s true, then the three rules of understanding data have to be “context, context, context.” Context gives meaning and insight to the data. Without it, data is just letters and numbers in rows and columns. Here’s what’s interesting: numbers alone provide no visual context. It’s best to show raw numbers to represent a single point of data. Here’s an example: 32. Is that a lot or a little? It’s a pretty good ACT score, but an NBA team with 32 points at the end of a game is going to lose. Without context, a number is just a number, and on its own is rather meaningless.
Unfortunately, showing a bunch of raw numbers together (think of our crosstab from earlier) doesn’t create quickly consumable context, either. That’s why when we look at this from a visual design perspective, if (and that’s a big “if”) we’re going to show raw numbers, they should be rolled up to the highest level and be used to highlight only the most important parts of the data. The example below highlights four key metrics of data and uses visualizations to provide details and context for those key metrics:
Create context with visualizations
Now that we’ve talked about when and how to use raw numbers, let’s circle back and look at the strengths of visualizations. It’s only when groups of numbers are thoughtfully brought together (often through aggregation) to create a visualization that we begin to understand the story the data is telling. Where numbers inform with precision, visualizations create context and evoke feelings. When I look at the map of “Usage by City” above, it feels like Seattle, New York, and San Francisco are big users of the site, and I don’t need to know the exact numbers to feel confident with that analysis. I see the visualization, and I quickly understand the story of the data.
Selecting the right visualization for your data
While it might seem like the easy way to help users understand your data is to shift from crosstabs to visualizations, it’s not quite so simple. There is a smorgasbord of visualization options to choose from. Again, the goal here is to select a visualization that shows the relationship you want the end user to understand. Comparisons? Relationships? Distributions? This gets into the best practices of visual design: a space ruled by the likes of Stephen Few and Edward Tufte, to name a few of the great minds in the visual analytics world. But for those looking for a simple way to make better data visualization decisions, check out Data Viz Catalogue. It’s one of the many excellent online tools to help you get going in the right direction to let your data tell the story.
Knowing when and how to use numbers and visualizations to tell the story of your data is key to creating rapid insights, understanding, and ultimately making better data-driven decisions. By following some of these best practices, your data will be more consumable and dramatically increase the overall user experience.
Nelson Davis helps lead Slalom’s Tableau team in Atlanta and is a very active member of the Atlanta Tableau Users Group. Nelson is always looking for opportunities to encourage others to take their Tableau work to the next level.