This example was the *winner* of the 2005 DM Review magazine Data Visualization Contest, Scenario 3 (dashboard)
Click here to see the SAS code.
Since the 'target' values were provided by quarter, I decided to summarize most of the data by quarter. I generate 17 bar charts showing various quarterly metrics, and I overlay a special pointer showing the target. I color-code these 17 charts consistently, showing the good/satisfactory/poor performance.
The response axes on each chart start at zero, and have a number of tickmarks/divisions that best suits the data in that chart (rather than scaling them all the same). Since "on time" values could go to 100% the axis is scaled to 100%, since satisfaction values have a max of 5 the axis has a max value of 5, and the other charts are scaled up to the next 'nice' number that accommodates both the maximum bar height and the target value (whichever is higher).
In the sales pipeline chart, I ordered the bar slices by highest probability to lowest (with highest at the zero baseline), and shaded them so that the highest probability sales were the darkest. This produces an optical effect that reinforces the data -- the lightest color segments at the end of the bars are the sales with least probability of happening.
In the Top 10 Customers bar chart, the blue segment of the bar represents the year-to-date (YTD) sales, and the light segment represents sales that are in the pipeline. I use a line in the blue segment to show the quarter-to-date (QTD) sales as a part of the YTD sales. I created a custom bar segment legend to show how the blue bar segment represents both the YTD and QTD values. (YTD and QTD values were given in the spreadsheet, and I subtracted the two to get the size of the YTD-QTD bar segment correct).
The software I used generates an html overlay, which provides html charttips/flyover-text, so when you hover your mouse over the bars, you can see the exact/full values for that bar. I've also set up html href drilldowns so you can click on the bars and that will bring up the spreadsheet (I could have alternatively had it drilldown to a table, report, or a more detailed graph). These capabilities are very desirable in a dashboard.
Also, the way I set up this dashboard, I wrote a program to read the blocks of data directly out of the spreadsheet, and transposed/summarized/joined the data in various ways to get the data I plotted -- having done this programmatically, the code can easily be re-used with other/similar data.