I caught an interesting post over on the Juice Analytics blog regarding Dashboards. They were arguing against the common wisdom of defining a Dashboard as a ‘single page of information’.
I’d have to say I agree. It seems an entirely arbitrary rule for the display of information. I’ve seen very usable and insightful dashboards that are multiple pages long. They also note that the idea of a space restriction is only one of several ways to enforce constraining information to what is valuable and useful – another way is simply restricting the number of measures. If only people would do that!
While the whole discussion on what is or isn’t a Dashboard is interesting, it’s less interesting in my book than understanding information typologies that are used in decision making. Dashboards are an end point in this process, not (as they are typically made out to be) a jump-off point.
Information types for businesses fall into two major categories:
1. Actionable – information where the root cause of a change in state is known. On a car Dashboard this would be the fuel gauge – you know exactly what to do when it hits empty.
2. Derivative – information that you know is important to understand and track, but the measure itself is derived from multiple areas and therefore state changes could reflect a multitude of underlying issues. In the car example, this is the ‘fix engine’ light – very useful to know, impossible to understand.
If you think of Dashboards as displaying either one or both of these types of information, the design you employ is driven by the information typology, not any particular arbitrary layout rule.
A Dashboard that is heavy in Actionable information needs to measure the point of action. It’s no good having the ability to drill down to region when you need to know ‘at a glance’ if all your regions are ‘full on fuel’. This could arguable dictate a high degree of detail in a single layout.
A Dashboard that has Derivative information on it needs more interaction. If something at an aggregated level is trending down, you need to be able to drill into the reasons why. Hopefully the business case for putting together a heavily Derivative Dashboard would have included a ‘causation pathway’ for you to investigate.
There are probably other information typologies you could add to the list. There are certainly subsets of the two I mentioned here (Derivative measures could have underlying quantitative and/or qualitative issues). but by and large, it seems a better starting point for figuring out (and even defining) the essence of ‘Dashboarding’ than whether or not information should be on a single page.
An while we are on the subject, why hasn’t anyone come up with a better solution to the ‘check engine’ light indicator? In this day and age, you would expect some of these fancy cars to know what the actual problem is!
UPDATE: Just saw a recent post over on Tim Ferris blog that covers some of this from guest blogger Eric Ries, co-founder and CTO of IMVU.
