Registered: 1183040614 Posts: 60
Reply with quote #1
I've put together a dashboard ranking my company's top 25 salespeople. In the dashboard I'd like to include a microchart depicting the prior monthly rankings for each salesperson. I'm currently displaying the prior rankings relative to the salesperson's current month ranking. For example, if salesperson A is ranked 5 this month, and was 5 last month, 6 the month before that, and 3 the month before that, I want my graphic to show those changes in a way that the reader can not only see the change in rank, but also know the salesperson's actual prior rankings. Ideally, I'd like to show the last 12+ months rankings for each salesperson. I've used microbarcharts and sparklines to visualize the rankings, but both methods provide graphs that aren't intuitive enough for explaining changes in rank. I've also tried win/lose charts, but they don't encode the relative rank of the salesperson, which is something I think is important. I've attached a snippet of my dashboard showing the current rank, and the microbarcharts showing the prior 14 months rankings, the right most bar represents the current month's rankings. Can anyone come up with something simple and compact (will fit in a standard width Excel cell) for this? Thanks. Attached Images
Registered: 1136226912 Posts: 245
Reply with quote #2
Assuming the bars are in chronological order, left-to-right, I'd put the "current rank" number on the right-hand-side of the charts.
Registered: 1182202224 Posts: 97
Reply with quote #3
I agree with grasshopper that the current rank should be on the right with the oldest measure on the left, as this matches the way we tend to view time (proceeding from left to right).
If you need exact values, you're in a bit of a pickle. Graphs, by their nature, aren't designed to display exact values, and small graphs like sparklines definitely don't work well for this (sparklines don't even have their quantitative scales labeled). If exact values are necessary, displaying the raw numbers is probably your only option. Hopefully, you can live without exact values. There are a few things you can do to help people maintain context when looking at your graphs. First, I would use lines instead of bars, because they will highlight the overall change better and they will probably be easier to look at than hundreds of tiny bars. Second, I would make the fill color in the plot area of each of the graphs a little more salient so that, although the scale is not labeled, it's easy to see where the scale begins and ends and to see roughly where a particular value falls along it. Finally, you might want to highlight the current value in the sparklines, perhaps by putting a small dot on the right end of the line. This will help highlight the most recent value and give people a point of reference to compare the rest of the line to. Also, if possible, perhaps people could get the exact values by hovering their mouse over the month of interest? People won't be able to tell whether a salesperson came in 6th or 7th three months ago, but it will be obvious they didn't come in 15th, and the overall trends and exceptions, which are probably more important, will be obvious. -Bryan
Registered: 1155665203 Posts: 69
Reply with quote #4
If you really need something highly compact, yet giving a reasonably exact number out of 25, I would suggest six thin gray lines, one pixel high, with four white pixels between each. This defines twenty six possible levels from highest to lowest.
Then, for each time period, have a single pixel, or some block an odd number of pixels high, be placed within these "musical staves". It should be possible to count off the staves and see where each pixel is in relation to the two lines it's between, to arrive at a fairly precise figure, all within a line less than thirty pixels high!
With many modern display devices I suspect this will simply be too small to see, and the dimensions will have to be doubled up for clarity. Also, this technique partly depends on having a display technology of either a predictable number of pixels, or many more pixels than necessary, otherwise there'll be unpleasant visual effects from pixel quantisation.
By the way, speaking of unpleasant visual effects, your microcharts picture doesn't have to be JPEG, which shows the characteristic "inky fingers" effect on tect and graphics, and is almost always less compact than a GIF or PNG would have been anyway. Attached Images