After you have defined the business purposes and you know what metrics you’re looking for, take the time to build a dashboard. The investment will be well worth it.
Imagine if your data consisted of a plain table with a time variable and amount of sales. You could present this in a nice excel table, but certainly you would present the time series data as a line graph. Instantly the audience would receive this information and visually identify trends or local maxima and minima. Visual presentation creates instant recognition of how the underlying data behaves in a way a plain table can not. Similarly dashboards combine multiple metrics and compile them into a simple visual reporting format that communicates more efficiently than pages and pages of boring tables. Dashboards are especially helpful when presenting data to managers who are interested in only what is most important to a decision.
There are three keys to keep in mind when choosing the appropriate representation of your data. First, what is underlying data? This seems obvious but there are many examples of poor choices of presented data. With regards to the image below, a response for one candidate did not prevent the respondent from also responding for another candidates. Answers were not mutually exclusive. In this example the survey was not done correctly, but it should be obvious that data represented in a pie chart should be percentages of the total, and the total equals 100%.
Secondly, keep it simple. The point of a graphic is to summarize. The graphic is not a narrative. When you look at your speedometer, you don’t want a narrative about how you slowed down when you passed the bicyclist and sped up at the yellow light. You want to know current speed in relation to acceptable norms. Speed dials are a great way to look at current KPI’s in the context of desired ranges.
You don’t need to know anything else about this product, but sales are not where they should be. Sales are close, but can be improved upon. The graphic instantly communicates what is vital. It answers the how are we doing questions at an aggregate level. To answer why questions more granularly a deeper exploration is necessary.
Lastly, know your audience. If you are the end user of your dashboard, then things are easy. But if someone else will be using your dashboard to monitor the state of things, make sure it is clear what is going on. Use titles that are informative, label your axis or metrics. Color is a great way to distinguish between categories. As indicated above red, yellow and green instantly convey what is good and bad when used appropriately.
Dashboarding is a great opportunity to add value by combining creativity and analytical savvy. A good practice is to gain feedback from peers and see if they can quickly interpret what you are trying to communicate. If not go back to the drawing board.
There is a growing host of software that assists in creating dashboards. Google analytics has a built in dashboarding feature. Also there is Tableau, Infocaptor, and Intelliview to name a few.
Here are a couple of articles and blogs that have helped me think of new ways to present data:
http://flowingdata.com/ - (This might be my favorite)
http://www.information-management.com/news/10001129-1.html - (Best Practices) http://fivethirtyeight.blogs.nytimes.com/ - (Not about dashboarding, but good example of effective use)