Dashboards Done Right: Insurance Dashboard by Ellen Blackburn
My friend Andy Kriebel has a website Data Viz Done Right where he talks about data visualization best practice by showing examples he has found on the internet. I thought I would do a similar post focusing on dashboards. Ellen Blackburn recently posted this insurance dashboard. This is really well done. In fact, it incorporates so many of the things I teach in my data visualization classes and workshops and discussed in The Big Book of Dashboards. So here is a short write up discussing some of these things.
About Ellen Blackburn - After finishing her studies in applied social and biological sciences, Ellen joined the Information Lab’s Data School, where she currently works as a Consulting Analyst. In addition to her day job, Ellen utilizes both Tableau and a long-standing interest in design to highlight important topics through compelling data stories.
There are number of things that stand out to me that make this a great dashboard:
Well Organized - The dashboard is organized well, with the setting menu on the left-hand side and navigation in the top right corner.
Designed to a grid - The dashboard is not only organized well, but the placement of the elements on the dashboard is designed to a grid. It is a basic four quadrant dashboard with BANs across the top. The horizontal and vertical lines show the very clear grid design.
The use of BANs - The big numbers are placed across the top of the dashboard as the "headline". This makes it easy to see the important metrics, scanning right across the top of the dashboard.
Additional context - In addition to the BANs, there is additional context to the numbers, showing the prior year numbers and an indicator for easy comparison.
The Use of color - The use of color in this dashboard is terrific. There are two colors showing good vs. bad, a dark blue for increases year-over-year and an orange color for decrease. The use of blue and orange is colorblind friendly, so users with color vision deficiency will be able to see the difference. In addition, there are icons used as indicators, making it even easier to spot the increases vs. the decreases. There is also a light blue highlight color used to highlight the selected item.
Chart types for the various comparisons - The use of bar charts, diverging bar charts, and stacked bar charts make the comparisons easy to see.
Good use of a stacked bar chart - Stacked bar charts that are stacked with multiple categories can make comparisons very hard. However, in this case, the light blue portion of the bar shows how much the selected item has out of the top N.
Tooltip design - The custom tooltip design is very clean and easy to read. It provides additional data without clutter.
Font Contrast - The contrast of font is handled with great care. The titles include all caps in Bold, and while there are two different font sizes in the title, it works very well. This style continues into the chart titles and BANs. Then a lower level font is used for labels and other text.
Use of Top N dropdown - To avoid lots of scrolling on the dashboard, there is a Top N dropdown box. Selecting Top 10 avoids scrolling on the dashboard and creates a very clean layout. If the user needs to see more, then changing it to Top 20 shows all of them and adds the scroll on the charts.
The information icon in the top righ-hand corner provides a great description and formulas for the acyronyms on the dashboard.
The author and data source are listed on the visualization. This is always a good practice to make sure readers know who created the dashboard and where the data came from. While this may not be necessary if publishing on your Tableau Server at the office, it is very important when publishing in the wild.
Iterate. Iterate. Iterate. - It might not be obvious from looking at the dashboard, but there was a good bit of iteration on this dashboard. Ellen made lots of little changes along the way, and even when I reached out to her, on what I thought might be a finished dashboard, she was still making changes and tweaks.
After reaching out to Ellen about this dashboard, she made some additional changes, so there is not much else I would do. Below are some additional things that might be considered.
The bar charts are so clean that it's almost too clean. There is a single label on the top bar and there are no axis rulers or labels. Therefore, the user will be required to use the tooltip in an interactive format to see the value of any bar other than the highlighted bar, or they will have to use the single label on the highlighted bar to make an estimate of the other bar values (which is scaled as a %).
Great care was taken in the sorting of the bar charts on the bottom. The selected item sorts to the top, while the rest of the bars are alphabetical order. This makes sense, given there are different ranks based on the 3 columns, so it would be impossible to sort them all with consistent labeling. A future enhancement might be the ability to sort each of the three columns by rank (and maybe adding the rank to the tooltip).
Stylistically, I generally prefer not to use so many divider lines on my dashboards. This dashboard has a very strong grid, so adding so many additional divider lines aren't really necessary. However, the choices that were made here work really well to reduce the effect of these lines. In fact, I don't really mind them on this dashboard because of the way they are done. The divider lines are thin, muted in a light gray and are not connected. These design choices keep the dashboard from looking cluttered or the feeling that everything is boxed in.