Thanks for your clear discussion of the merits of the funnel analogue. I can see how a logically suspect but well understood concept can be useful for conveying information. There must be dozens of similar metaphors, which defy logic but still enhance comprehension. I should remember the engineer’s credo:

If it works, don’t fix it.

Not the other engineer’s credo:

If it works, you haven’t fiddled with it enough.

I’ve fiddled with Chandoo’s funnel chart, and made it work for technical folks like myself, but perhaps broken it for those who market and sell what the technical people produce.

And Chandoo, Tim, and Rene: I’ll try to keep an open mind, even while spreading what I think are best practices.

From a purely logical perspective, I’d agree with you. A bar chart with all bars starting from a common baseline provides as much information as the funnel chart in about half the space.

However, from my experience with sales people, for whom “sales funnel” is such a basic term, the funnel chart is a common representation of the fact that the number of prospects declines the further down the sales process one goes. The advantage of the funnel metaphor in this case is that they understand it already. If you show a funnel chart to sales executives, you wouldn’t have to explain it.

Bar charts with common baselines don’t imply a “mandatory narrowing” — it is always possible that the order of the bars may not be sorted in accordance with their value. Thus, a sales executive may spend a few seconds to try and validate the bar chart mentally, something he wouldn’t do with a funnel chart.

If the goal of a dashboard is to deliver information that the audience can grasp in as little time as possible, then the funnel chart, despite its waste of space, may be the better choice for audiences who are used to the funnel metaphor.

I agree. The funnel is not such a good metaphor. And the baseline common baseline of bars in a regular bar chart make it easier to compare one phase to another.