Dashboard design

Dashboard design is an increasingly popular topic within business intelligence. Thus, I have dedicated a special page for best practices to help you achieve the best dashboard design.

What defines good dashboard design?

I believe most people implicitly have an understand of "good" as either something functional or something visual. I will here define "good" as "effective", i.e. a good dashboard is effective in:

1. communicating its information to the user
2. motivating the user to take concrete action on the presented information
3. allowing the user to explore data for further information/details

Functional requirements
First, let's define the purpose and goals of a business intelligence dashboard (aka Management Information Dashboard):

A dashboard must supply the user a combined view of all the most important data he/she needs to carry out tasks within a certain context in a presentation that is simple to operate and easy to read.

From this definition we infer that a Dashoard:

1. must be role-based
2. must be context-based
3. must fit inside one screen (no scrolling)
4. must not require the user to enter parameters before showing data.
5. must provide interpretation of the data, i.e. rules and benchmarks, that allow the user to quickly judge whether the presented KPIs are "good" or "bad" thus leading to a relevant action
6. must be interactive so the user can extract further information from the displayed data on demand

Visual requirements

The visual design is important because a dashboard must look appealing - rather than embarrassing - to entice the user to use it. There are many theories about visual design - in fact this topic has far greater implications than just dashboard design. So consider the following advises as merely a starting point for your journey into the optimal dashboard design.

1. One screen. Design the dasboard to be no larger than what fits on one screen. No scrolling must be required. The big challenge here is limitation, i.e. don't overload the dashboard - question yourself again and again: What is the most important? I know I already mentioned this under functional requirements, but it is both a functional and a visual requirement.
2. Sense of overview. Look at your dashboard from a distance and ask yourself a couple of questions: Does it give me a sense of overview? Can I quickly find certain objects?
3. Acclimatization. Embrace the environment that the dashboard will be presented in. Make due observance to company standards. Have a look at the logo, use of colors etc. Also consider the recipient: What are their expectations? What kinds of visualizations are natural for them?
4. Use proper highlighting. A gauge full of vivid colors may look appealing when viewed alone. However, in a dashboard with multiple visual objects the use of extensive highlighting usually has the opposite effect: It makes the viewer lose sight of the important details; it all becomes a jumble of colors and effects. Spare the cool effects for what really needs to be highlighted.
5. Use proper visualization. You probably have an arsenal of grids, gauges, graphs and maybe even maps at your disposal. Be careful to use proper elements for the data you are representing. For example, gauges are good for relative measures (percentages) but not for absolute measures (e.g. sales amount). Area graphs are good for displaying development in stocks. Grids are good for showing exact numbers, but not for showing trends in time.