Data Visualization: Three Best Practices

When done right, data visualization gives us a clear idea of what the information means by giving it visual context. In particular, it allows us to identify trends and patterns that might be hard to detect or fully appreciate in a tabular or other non-visual format. This is the value proposition of business intelligence (BI) tools such as Looker, Tableau and Chartio.

At the Modern Data Stack 2020 Conference, organized by Fivetran, Kate Srachnyi, founder of the consultancy Story by Data, offered her best advice for taking your data visualizations to the next level.

“My main message for everybody is this: Do not accept the default settings,” said Srachnyi. “Take the steps necessary to enhance your data visualizations to effectively tell your data story.”Below are three of her top recommendations for creating visualizations that are informative, appealing and efficient:

1. Design with intentionality

Design with your audience in mind — are people going to be reading it in a brochure, on their mobile phone, or printed out at home? Is it static or will they interact with it? If the latter, is it easy to understand where to touch, tap or filter?It’s important that your visualization highlights the most important takeaway its intended to convey.

“One way to test this is to just give it to someone who has never seen your dashboard before and have them use it,” said Srachnyi. Watch them as they interact with what you’ve created, and if they get a confused look on their face, you know you need to adjust.”

2. Reduce clutter wherever possible

Removing unnecessary elements from the view can significantly improve its quality and impact. “Less is more,” Srachnyi declared. “What tends to happen when we get our hands on these data visualization tools is that we can visualize very easily and very quickly. So, we put 15 charts in a dashboard using all these colors and text and images. And things start to get cluttered. So, what we need to do is be very intentional.”

For example, she said, cut out any icons that don’t support the story you’re trying to tell. Use data legends sparingly and make sure they’re placed near the data they’re referring to. Remove unnecessary grid lines or borders (or make them a light gray).

3. Use color intentionally

Poorly choosing colors will leave your audience distracted or confused. Instead, use color to guide your audience’s attention to the patterns and data points that are most important for telling your story.

Use consistent colors for consistent variables, and present supporting data in muted tones like gray. Stick to five colors or fewer overall. Make sure that patterns are still viewable in grayscale, in case people print it out on their home printer (which is far more common now that we’re all working from home). Various online tools can also help ensure people with red-green colorblindness can still easily interpret your visualization.

Take a second look to ensure effect

The additional effort you put in will show through in the end product, said Srachnyi. “Spend an extra 15 minutes playing with the settings. There are so many features and options that can take your data visualization from just looking good to being really great and effective at telling that data story.”