Cursory Dashboard Principles

Last week I received an email from Bernard Lebelle, who is apparently updating his book on dashboarding in Excel and will be including screen shots of the three winning tools from Chandoo’s Excel Salary Contest.   You can download my submission here. Mr. Lebelle’s book is in French so I can’t speak to its quality, but my hope is that it will espouse good dashboard design principles that make knowledge transfer and decision making easier.

I was asked to send “a few lines presenting the design principles behind your creation.”  I’m not sure how much of my response will make the book, and given the book is in French, I may never know, so I thought I’d post it here.  The points are by no means exhaustive, or even prioritized, but all are worth considering the next time you are putting together a dashboard or analytic tool.

The data should be the star of the show.  There are no distracting extra elements like clip art, background images or grids (chart junk as Tufte would call it).

Don’t distort the data. All the bar chart axes start at zero, so comparing bar lengths is meaningful. No exploding pies, no 3D graphs.

Use color judiciously to highlight and provide meaning.  There is only one color other than black/grey/white and it is used to show the currently selected data and set it apart from what it is compared to.  This consistent and restrained color usage draws attention where it should be directed and gives the dashboard a professional look.

Alignment is important.  The elements are all aligned to a grid to give it a pleasing, organized layout and an easy to follow flow.

Provide context. In this case the context of the whole data set is always available for comparison against the selected data set.

Put it on one screen. Information seen together on one screen allows you to identify relationships and make connections.  You can pack a lot of information in one screen without losing meaning or depth.

Keep interaction simple and elegant.This dashboard has only one interaction: filtering by region and/or position.  It is easy to understand and achieve with all the instruction necessary summed up in a single sentence.

Make it responsive.  Response to user interactions should be immediate (or at least feel immediate).  If people have to wait for the computer they will be frustrated with the tool and/or they will spend less time gathering insight and more time staring at the “calculating %”.

To see another quality entry to the contest which, sadly, didn’t fare as well as I’d hoped in the online voting system, you can view Ben Jones’ write-up about his submission.

This entry was posted in Analytics, Visualization. Bookmark the permalink.

One Response to Cursory Dashboard Principles

  1. Ben Jones says:

    Hi Joey – thanks for the reference to my entry! I had high hopes for that entry, mostly because my kids really wanted the X-Box. ;)

    I’m a big fan of 1) the simple color palette and 2) the dual histogram charts on your entry. Using few colors and column widths to effectively show two different distributions in one graph was very effective.

    Thanks again, I hope the French appreciate your response. :)

    Ben