Take a look at the innovation heat map published by McKinsey (link to article)
I call this the “un-innovative heat map on innovation” for obvious reasons.
First let me fill you on what the chart is trying to show. Folks at McK got curious to know how innovation around the world is and partnered with World Economic Forum to create this heat map. The map shows innovation clusters (sized based on the number of patents won) on 2 dimensions – Diversity (how many companies & patent sectors are in the cluster) and Momentum (growth rate of patents). If you are wondering what a cluster is, it is a city.
Well, Mckinsey folks are always keen to plot anything including your cat on two dimensions. So they created a gazillion bubbles and plotted them on 2 axes and conveniently sliced the area in to 9 parts and named them like hot springs, shrinking pools, molten lava (well, not really, but the first two are true, I swear!!!)
One of the primary shortfalls of this heat map is, it takes innovation clusters (cities) that already have geographical co-ordinates and plots them in a way that is unreadable.
How could they have improved this heat map?
Instead of plotting the bubbles on 2 dimensions like diversity and momentum, they should have used simple google maps API or Many eyes world map visualization and colored the bubbles based on whether the cluster is in a hot spring or a stinking ooops, shrinking pool. That would have improved the effectiveness of this heat map so much more. Hey, it also helps you locate your state or city easily.
Your comments on this heat map – Hot or not?
Previously on infographic inspiration: Bubble Chart Fail, Bubble Chart Success, Kiss and Impress, More visualization inspiration
8 Responses to “Hot or Not – McKinsey’s Innovation Heatmap”
That's a bubble chart, not a heat map.
One thing I notice is that the big bubbles are on the "diverse" right. Well, of course they are: all else being equal, bigger areas are likely to have more diversity. McKinsey ought to have normalised the diversity to take account of the size, perhaps using measures developed in ecology.
Good point. Number of patents per company would work better for these scales.
The chart is impossible to read. Perhaps unfilled circles would help to show all of the bubbles. And why single out SF and Mpls when underneath there's a much larger bubble, larger than Tokyo. Calling the size axis Diversity of Cluster is a PC misnomer: it's Size of City and should be labeled as such.
But these consulting companies don't make bucks for saying anything in a clear and obvious way, do they?
This is typical of McKinsey work - generally useless, confusing and designed to provide potential clients with the implied message that they are the only ones sophisticated enough to "understand" the junk they publish. I've had the chance to see some of their "work" and it is NOT impressive.
This is not the kind of chart that is worth publishing on a site like this.
It is quite confusing. I do see what it refers to but it is very lousy. Thanks for the link to many eyes - it is a blast!
Would be interesting to know what the folks over at World Economic Forum infers out of this chart with no story to tell. 🙂
I agree, this chart is quite pretty, but getting any real meaning out of the chart is very difficult.
There might be 200 bubbles on this chart, but only 21 can be positively identified. If they had labelled the largest 21, that might make sense, but as JP pointed out, the second largest bubble is un-labelled, along with many other large ones. If they had labelled the 21 nearest to the right, that might also make sense. But they haven't. If they had labelled the 21 largest US cities, that might make sense. But they haven't. If they had labelled the largest city in each of the top 21 countries, that might make sense. But they haven't.
If the message was "Silicon Valley and Tokyo are far ahead of everywhere else" then they could have easily left off the smallest 150 bubbles.
I understand that the size of the bubble is related to number of patents. So, how about a scale - the legend in the top right should say something like "=100 patents granted in 2008" or whatever the number is.
I understand the vertical axis. But the horizontal axis is quite difficult to understand, and could conceivably have been calculated in several different ways, and therefore could have been open to manipulation. For example, if a cluster has one company working in one sector, and a second company working in that same sector plus one more sector, what value is returned ? 4 ? 5 ? some other number ? It wouldn't hurt to show the actual values for a couple of example bubbles - for example Silicon Valley and Stockholm. And what about patents that are not granted to companies, for example to individuals, or to educational institutions ?
Actually, the size of the bubble doesn’t reflect the size of the city, but the number of patents filed by companies or inventors in that city as well as towns or suburbs in its surrounding area. For example, Dayton, OH (pop. ~147, 000) represents 448 patents while the larger city of Toledo, OH (pop. ~286, 000) has only 256 – and you can see Toledo’s bubble is smaller.
I agree that charting innovation clusters on a world map is another interesting option for displaying the data, making it easy for a user to locate them by region or search for specific cities. However, their geographical coordinates are probably not the most interesting fact here. The rate and diversity of growth are the more compelling metrics to map, as this allows us to easily compare the performance of geographically disparate clusters. We see, for example, a fairly high number of patents filed (momentum) in the Bristol, UK cluster from 1997 to 2006. These are patents from fewer companies or patent sectors (diversity) than are evident in the Silicon Valley cluster.
We’ve just updated the chart with an interactive version. The introduction speaks a bit more about the data. http://whatmatters.mckinseydigital.com/flash/innovation_clusters/ You can now identify all the cities, which raises interesting questions—for example, can the Fort Collins, CO cluster, ranked surprisingly high in momentum, continue to grow in the number of companies and sectors actively innovating and thereby move into the same deep waters as the Taipei, Taiwan cluster?
It’s difficult to judge how much to clutter a chart like this with complex explanations of the methodology. But I can see that we may have presented this with too few words.
@Derek: that is not my word. McK choose to call this a heat map. And good point on the bubble sizes. Have you seen Mary's response further below ?
@Angelo: I am sure folks at McK would disagree with you. As far as I go, they actively contribute to world knowledge and analytics (although sometimes it results in confusion, when you try to makesense out of everything)...
@Alan: Manyeyes is an excellent visualization platform, that has given us stuff like tag-clouds, several interactive visualizations. It takes the labor out of making visualizations...
@Gerald: aah, the labeling... at least the new interactive version (link: http://whatmatters.mckinseydigital.com/flash/innovation_clusters/) is somewhat better with the labels. But it still has the problem that first you need to hover on 100 points to findout the city you want...
@Mary: thanks for commenting, and welcome to PHD. I agree that your intention is not to show the geographical co-ordinates of the cities, but that is never really information. It is context.
and your new interactive version makes it easier to read the chart.