I want to talk a little about bullet graphs today. For those of you who don’t know, a bullet graph is a type of column/bar graph developed by visualization expert Stephen Few to serve as a replacement for dashboard gauges and meters. Stephen developed bullet graphs to allow for the clear display of multiple layers of information without occupying a lot of space on a dashboard.
.
A bullet graph contains single performance measure and compares that measure to a target. A full blown Bullet Graph will also show the measure in the context of qualitative ranges, such as Poor, Fair, Good, and Very Good.
.
.
If you’re thinking that this bullet graph looks suspiciously like a glorified thermometer chart, I agree (sort of). In his design though, Stephen did add an extra dimension of the colored qualitative ranges, showing a ton of data in a very compact space. This comes in handy in a dashboard environment where screen real-estate really counts. So in the end, Stephen did come up with a relatively new way of consolidating lots of information in a graphical form.
.
But here is the problem with Bullet Graphs in general. It has been my experience that I can’t use this full blown version of bullet graphs very often. Each time I have tried to use a bullet graph (complete with the colored qualitative ranges) I have to take extraordinary efforts to explain what the graph means. There is a lot going on in a bullet graph, and managers just aren’t used to seeing them. Ironically, packing so much information into such a compact space actually becomes a barrier to understanding the data.
.
I can remove the qualitative bands so managers pick up the meaning right away – no explanation needed. However, that would essentially bring me back to thermometer charts. You’re probably thinking that I should work with a better class of managers. I can’t agree more, but for other reasons.
.
The reality is that I have found it difficult to implement the full blown version of bullet graphs without a training session on how to read them. And in the corporate environment, this is rarely possible.
.
So how about it? Have you found Bullet Graphs usable?
.
Creating Bullet Graphs in Excel
For those of you who want to take a shot at creating bullet graphs with Excel, I came up with a method while writing “Excel Dashboards and Reports For Dummies. With my method, I can do things like create multiple bullet graphs on one chart and give the illusion of solid bands across multiple measures.
.
To see my method, you can buy the book (wink wink).
For all you cheap bastards out there, I’ve got a free video that demonstrates my method.
.
By the way, Charley Kyd has also developed a method of creating a bullet graph in Excel. But seeing that I’m not as smart as Charley, I found his method to be a bit too involved for me.
Mike, I’m actually toying with the idea of submitting a tip through the Microsoft Excel blog (since I have no other outlet for now) for creating bullet graphs in Excel 2007 and higher. The technique does not use column or bar graphs for qualitative bands (hence the reason why it won’t work with earlier versions of Excel).
I can’t imagine anyone understanding the qualitative bands unless they are based on established standards within the organization for the measures involved (a point I’ve noted). Otherwise, you’re better off without them.
I’ve found the bullet charts to be less effective than I would have thought. Partly it’s related to Mike’s observation that you have to explain them every time you use them. Partly it takes a while to sort out the gray bands. I guess here’s one case where red-amber-green may actually help, buth by reducing explanations and by speeding up comprehension..
Mike’s solid bands option is a reasonable compromise, but it still takes a little extra effort to read, and it assumes the levels of the bands are the same for all categories.
here is a simple method to create a horizontal bullet graph in excel 2007 via the error bar :
http://excelpro.blog.sohu.com/131084696.html
If I had to choose between a bullet graph and a sparkline with a quantitative band (i.e. a sparkline with a horizontal band denoting some kind of ‘acceptable range’), I’d probably pick the sparkline. This is because then you get to see how the thing you are tracking has changed over time. Bullet graphs only show what’s happenning right now.
That said, bullet graphs have a cooler name than sparklines.Try saying “I’m gonna pop a bullet graph in your ass” and you’ll hear what I mean.
If I had to choose between a big dial, and a bullet graph, then I’d choose the bullet graph.
On using multiple bullet graphs on one chart, when I asked Stephen Few about it he said “When all of the measures share the same quantitative scale and the same qualitative ranges (for example, the same ranges of good, satisfactory, and poor performance), there is no real reason to use bullet graphs. I think it would usually work best to use a single bar graph with continuous qualitative fill colors in the background. This would make it obvious that the measures belong together and can be compared in all respects.”
However, if the quantitive scales differ, then bullet graphs might be a good way to report. For example, in Few’s bullet graph design spec at http://www.perceptualedge.com/articles/misc/Bullet_Graph_Design_Spec.pdf you can see at the top of page 2, there’s 5 bullet graphs per example and each one has a different quantative scale (i.e. different target bands).
So using a solid band option wouldn’t work here.
Liu: Thanks for the link. Thank god for screenshots, especially on this site.
This has definately opened up a few ideas for me.
There must be something wrong with me. I had no problems understanding what the bullet graph was showing. Then again, I’m not a Manager.
DataPig, et. al.,
I’ve introduced the bullet graph to many individuals and groups over the last few years. What I’ve found is that it takes less than a minute to teach people how to read it. It takes a great deal more time to teach people how to read other useful forms of data display, such as box plots and scatter plots, but it’s worth the time when the graph can tell powerful stories. When Hans Rosling of GapMinder presents information to audiences throughout the world using an animated bubble chart, even people who might ordinarily run from scatter plots or bubble charts in horror become captivated because Rosling introduces the chart in a way that they can easily understand. The point that I’m trying to make is that the merits of a graph should not be judged by the audience’s ability to understand it without instruction. The first time William Playfair used a bar graph, he had to tell people how to read it. Merit ought to be judged by the graph’s ability to tell the story it is meant to tell–clearly, accurately, efficiently, and without undue effort by the viewer.
A series of background colors that vary by intensity from dark to light is an intuitive use of color for representing a quantitative or ordinal scale, such as poor, satisfactory, and good performance. People in countries that use red, yellow, and green traffic lights to convey these meanings don’t understand them because they are intuitive, but because they have learned these meanings. This took more effort to learn originally (although no one remembers the effort) than it takes to learn that the lighter the color is, the better the performance is.
The bullet graph was designed for a very specific purpose: to present a single measure of what’s going on, compared to something such as a target, within the context of qualitative states (for example, poor, satisfactory, and good performance). It was developed to present this information in a manner that allows people to scan a great deal of information on a dashboard to quickly get a sense of how things are going. The bullet graph is only one of many effective ways to display information on a dashboard. When other forms of display are better suited to the task, they should be used.
It’s been fun to observe the ways that the humble bullet graph has been successfully used to solve real problems and how it has evolved to solve additional problems that I hadn’t anticipated when I first introduced it. Discussions like the one we’re having here on this blog will help to make this happen. Discussions like this might also plant the seeds of entirely new forms of display that have yet to be developed.
Steve
Stephen: Damn you and your clear and well-thought-out remarks! You have an uncanny way of changing minds.
I do agree that merits of a graph should not be judged by the audience’s ability to understand it without instruction.
My only feeble push-back is that the corporate environment doesn’t allow for luxuries such as traing and think-time.
As I ponder your remarks, it occurs to me that in 5 years, this conversation will be moot. With the adoption of bullet graphs in a myriad of software (future versions of Xcelsius and Excel being the big ones in my world) , a large majority of business users will have enough exposure as to make them second nature.
Think of all the money you could have made if you had not given the design away. I guess the joke’s on you Mr. “well-thought-out remarks”