I had the opportunity last week to attend Stephen Few’s 3-day Visual Business Intelligence workshop in San Francisco. The material on data visualization and dashboard design was a good refresher as I had previously read his books on the topics. However, I did learn a couple of new things when I listened to some of Stephen’s stories regarding data visualization and dashboard consulting.
There were two major revelations I had as a result of the workshop: the true problem with speedometers (gauges) and why people feel they need beautified dashboards.
The real business problem behind speedometers
Yes, we’ve all heard that speedometers are used because they look cool and remind us of a car. We also know that they take up a lot of space and show very little data. But what if you need to fill up space on your dashboard? What’s the problem with having something like a speedometer to make the dashboard design feel complete? There isn’t a compelling reason why you shouldn’t use them in these scenarios.
The answer is that these scenarios should not exist. If you have too much real estate on the dashboard, odds are that you’re not displaying enough information. Many dashboards use tabs to break up information but most of the time, there’s no reason to do that - the use of tabs should be the exception rather than the norm. By breaking up your information into separate dashboards, it’s harder to correlate and analyze the information. Even if the information is not interdependent, why force an end user to jump from dashboard to dashboard if you don’t need to? Stephen mentioned in his workshop that tabbed dashboards are often the result of an inexperienced dashboard designer.
The damaging use of colors and data visualization choices
Stephen recognizes that dashboards should be beautiful. If the dashboard were ugly, no one would want to use it. There’s a reason why there’s a whole field dedicated to industrial design and user interface design. However, he points out that over-use of imagery and colors will make it harder for the dashboard user to easily spot important information. Consider the following two dashboard designs and see how long it takes you to answer these two questions: how many times did revenue not meet goal? What product has made the least revenue?
Click on the image to enlarge.
Click on the image to enlarge.
Note: assume the dashboard is showing ALL the most critical information to make a decision
After this test, it’s easy to see how important it is to be very conscious of the colors and data visualizations you choose. However, we still see a lot of dashboards that over-use colors and poorly chosen visualizations. Why is that?
It’s simple. The dashboard has no intrinsic value because the information displayed is not important. Dashboards are an office productivity tool similar to that of Excel or Word. Like with any productivity tool, you want your workspace to be clean so you can clearly see the numbers or words you’re working with. Now imagine if Excel had decorations around your numbers, used bright colors for cells by default, and added your company’s logo every time you started a spreadsheet. I’m sure you’d turn off all these colors and decorations to get it back to a clean spreadsheet. You would do this because you don’t want colors and loud images to distract you from doing your job when you’re trying to make sense of important numbers. In addition, you don't want the data visualizations used to make it harder to understand the data as demostrated with the pie chart example.
If you found value in your dashboard, you wouldn’t want to be distracted by unnecessary decorations most dashboards have. Some dashboards I see in the wild look like they were built by a teenager whose design experience consists of building MySpace pages.
Attending the workshop helped me identify the root causes of these poor design practices. If you have the chance, I highly recommend attending Stephen Few’s public workshop. It’s a great way to learn the fundamentals and to hear real world stories on dashboard design challenges. His workshop schedule can be found here.