The ZingChart team recently came across a copy of the eBook Data Driven: Creating a Data Culture by DJ Patil and Hilary Mason. We found some interesting points related to designing dataviz dashboards that are worth sharing.* *

The book begins with some interesting anecdotes about how different organizations have used dataviz to great business advantage.

The business case for data and dataviz

box for parcel delivery data viz
“FedEx and UPS are well known for using data to compete. UPS’s data led to the realization that, if its drivers took only right turns (limiting left turns), it would see a large improvement in fuel savings and safety, while reducing wasted time. The results were surprising: UPS shaved an astonishing 20.4 million miles off routes in a single year.”

That’s a powerful payoff on data analysis!

Here’s the dataviz part

flag with text
“The democratization of data is one of the most powerful ideas to come out of data science. Everyone in an organization should have access to as much data as legally possible.”

* *And making data visual can only help democratize it more, since visual data is usually easier to for humans to read than tables or CSV’s. Dashboards are a great way to do this.

Dataviz dashboard tips

Patil and Mason share five tips for designing dashboards that are good to keep in mind as your work on your next project.

1. Data vomit

dataviz example: data vomit
Sometimes less is more. Patil and Mason urge you not to add more and more items in an attempt to help the user. It removes the visual hierarchy of your dashboard and confuses the point. “Data vomit is bad and leads to frustration,” according to the authors. “The data becomes intimidating and, as a result, is just ignored.”

2. Time dependency

zooming and interval control in dataviz
Here’s a great test for whether or not you should include something in your dashboard: “Include data only if you know what you will do if something changes.” The most effective dashboards chart data at intervals that can show significant change. In the example above, the first chart is zoomed in too far, or showing update intervals too close together to make sense of any pattern or trend. The second chart is zoomed out/showing the intervals at a wider scale so you can clearly see the pattern.

Good dashboards also visualize data in the best format for what it represents. We all know it is difficult for human brains to make sense of too many pie slices. Pail and Mason’s recommendation? “Display the data in a form that allows action to be taken.”

If this sounds too difficult to do in a single dashboard, perhaps you need multiple dashboards, or a chooser/date picker for time intervals. For example, one dashboard might show data on an hourly scale, while another could show the daily view. Multiple dashboards can also help prevent data vomit.

3. Value

This item seems to be recommended as a periodic review of the value your dashboard provides. And, not surprisingly, the authors recommend updating the dashboard if it no longer provides value. This may seem like a no brainer, but it’s always a good to have a reminder that our work is not set in stone. This is especially important for organizations with evolving states of data infrastructure. Patil and Mason write, “As the organization’s data sophistication increases, it’s likely that old ways of measuring the system will become too simplistic. Hence, those older measures should be replaced with newer ones.”

4. Visual

styling example in dataviz
Visual styling is something we consider often at ZingChart, since our customers often request different features. But visual considerations go beyond the aesthetic. Usability is an important visual factor that can sometimes be overlooked for the sake of style. Some of the authors usability recommendations include:

5. Fatigue

too many alerts can create fatigue for dataviz users
Have you heard of “alert/alarm fatigue?” It is a term the authors use for the user experience when too many alarms and alerts are triggered by a dataviz. In dashboards with this problem, “The team becomes desensitized to the alerts, because they’re occurring so often and they’re frequently meaningless.” The suggested ways to avoid fatigue are:

  • Review alarms set on a dashboard, then consider the actions that should be taken once activated. Is an alarm really necessary for the triggered action?
  • Can the alerts be improved? False positives could be contributing to fatigue.
  • Remove an alert or an alarm if it does not serve a specific purpose or have a specific action that should be taken when triggered.

Custom dataviz and dashboards

mini dataviz dashboard example
When you’re ready to start making a dataviz dashboard of your own, we have a few tutorials that can help you set things up: