A chart in a recent, interesting McKinsey study made some weird data visualization choices. This post shows how to fix it in three steps.

If you’ve used the internet recently, you’ve probably seen the chart below popping up in your feed.

It’s from a recent McKinsey report on: “The Great Attrition is making hiring harder. Are you searching the right talent pools?”.

The report is about how companies can attract and retain talent by catering to five types of employees (“personas”). It’s based on a large survey of employees in six countries. One of the most interesting results is reported in this chart:

The report is interesting and beautifully prepared, but this chart has got me annoyed because it’s terrible.

I’m not talking survey methodology here: It’s the data visualization that’s off. That’s surprising because McKinsey is usually really good at this stuff*.*

Happily, it’s easy to fix in three steps, so I thought that would be an interesting thing to try!

Step 1: Change bubbles to bars

My first gripe with the chart is: Bubbles are hard to compare and they’re wrong choice for this data

Bubble charts are useful for 3D data (remember Hans Rosling’s amazing Ted talks?), but people struggle to compare their sizes.

To the authors’ credit, the values are the area** of the bubbles (not diameter)…. but this is 2D data (category x value) so why use bubbles at all?

Here’s the same chart using a bar chart.

That’s already an improvement because now we can easily compare the bars. However, the labels are small and hard to read, which takes us to Step 2.

Step 2: Make it vertical (better labels)

English text is written left-to-right, so it likes horizontal space. That’s my second gripe with the chart: The text boxes should be arranged vertically

Let’s see what that looks like by flipping the chart on its side.

Much better. Now can compare the bars AND read the labels.

Step 3: Shorten less important information

My final suggestion is twofold: (a) the last four categories are much smaller than the others. (b) some of the label texts seem redundant. In short, the chart has secondary information that could be shortened

Let’s shorten the labels and group the last four categories:

Erratum: As Amit Steiner helpfully pointed out, adding scores for a multi-select question as I’ve done here is, of course, utter nonsense. I’ll leave this section as-is for data visualisation purposes and as a reminder to myself to be more careful.

Admittedly there’s a trade-off here because we lose a bit of information. Nevertheless, in this case, I’d opt for clarity and readability.

You’ll also notice I cut the “other” bar. That’s generally considered a no-no, but I’ve always felt it’s ok for “other” categories, since you usually don’t compare them to anything else.

All in all, I find this chart much easier to parse than the other one. Nevertheless, I wasn’t quite finished so read on for two bonus steps below.

Shameless Plug: Curiosity

If you’ve made it this far, why not go ahead and give Curiosity a try?

Curiosity is a productivity app that gives you one place to search all your files and apps. That lets you save time and get more done.

Curiosity connects with the tools you already use, including your local folders and cloud apps like Google Drive or Slack. You can use the shortcut-powered launcher to open things quickly, and the file browser for deeper searches.

Unlike other search apps, Curiosity keeps your data safe on your computer and never sends it to the cloud.

Curiosity is available for free on Windows and Mac. You can also get a free two-week trial of Curiosity Pro (unlimited sources / search file contents).

Give your productivity a boost and download Curiosity.

Two Bonus Steps: Buckets and icons

There are two other things I wanted to try with this data, but I realize they don’t really fit in with McKinsey’s style, or the goal of the report.

Let’s have a look anyway, just for fun.

Buckets

There seem to be a couple of buckets for the reasons people leave. Now we’ve cleaned up the chart, we have space to add that dimension as well.

Here you can see immediately that Money / Career is still the dominant reason people quit, whereas Work-life balance still plays a subordinate role.

Icons

And while we’re doing colors, why not brighten things with icons instead of bars?

To be honest, I think the icons might be a bit much: They distract from the chart’s message without adding much… but it was worth a try!

It might work better if one used simpler icons and removed the “other” category.

And that’s a wrap

This was a quick exploration of how we could improve the dismal McKinsey chart from this interesting study.

I should add it’s understandable why the authors might have made the choices they did: Bar charts get boring after a while! And the fact it’s been shared so much certainly proves them right!

Nevertheless, I thought it would be fun to see how it would look “fixed” from a data visualisation perspective. I hope you had fun, too!

Here’s an instant-replay of the changes:

Let me know in the comments if you have any thoughts!