Bad BI Form

If you read Stephen Few’s work on BI Dashboards and visualization, you’ll recall that he says charts that rely on people to compare areas are not well suited to humans as we have trouble making the comparison.  Circles are especially bad, and forget about fancy donut (circles with a hole) shaped charts.

BI charts bad

This image is a perfect example of that.  Can you really compare the big circle with the others, or even the others to each other?  See how easy it is in the vertical bar comparison?  Keep this in mind when dealing with comparisons using area based visuals.





Posted on October 23, 2013, in BI Theory and Best Practices and tagged , , . Bookmark the permalink. 5 Comments.

  1. Nice concise post, Jeff, and good point. But here’s where I disagree with Mr. Few a bit: it depends on the observation/insight. In your example, you are totally right that observing that the US is outperforming the 10 other top countries by about 33% is definitely easier to see in the bar chart. However, if my observation that the US is strongly outperforming the next highest country (China), the bubble size works for me on quick glance. Now, if you took your stacked bar and made it a regular bar, it would work, too, but again, if the observation is that the other 10 countries are relatively equal in comparison to the US, the bubble works just as well for that. It just depends on the insight you’re driving towards.

    One thing that Few gets right is that you can use space more wisely on a dashboard with a bar chart, but his insistence that circular graphs are always better rendered as bars seems a bit too black and white for me.

    • Excellent point Kevin. If you are simply trying to show the outlier or an order of magnitude difference, then this visual certainly works.

      But to me it seems a bit too large for such a simple point. I’d hope that you could also get a secondary insight from it – something “one level down” in detail – perhaps a more accurate comparison between the countries. Maybe we might see that one of them is 2X the other – comparing circles will fail on that front. Also, as we know when we build a BI system, our data sets are never static; someone could change the filter criteria down to a scenario where the huge outlier is no longer there – then you are left with a bunch of circles that are hard to compare.

      Lots of ways to slice this stuff no doubt. I am always interested in the crazy charts that ESPN The Mag uses for sports statistics – some of them grab the attention but take a while to understand and then actually use. Here is an example of one – Crazy Chart Takes a while to even figure out what its trying to communicate, but its visually appealing no doubt. One has to find some sort of balance between visual impact and usability. Magazines and marketing materials can lean to the fancy visual more, but I think in BI we should be careful of that. Don’t want to be “all style and no substance”!

      Jeff M.

      • Wow…what a hot mess that graphic is! You’re right…it takes awhile to understand what it means.

        Your point about the outlier and changing data is sound, but your illustration depends on it. If you don’t have such a huge outlier like the US being more than the next top 10 countries combined, the bar visual will still fail. I can’t really make any sense of the stacked bar beyond that observation because there are too many slices with too many colors. You’re just trading the same size circles for the same size rectangles.

        That said, though, your overall message is a sound one and I like that you’re writing about it. Too often we don’t pay attention to what visualization we use and we don’t think about what happens when you apply a different filter or the data evolves over time. There are no easy answers and it bothers me that Few is so black and white about something that is very grey.

      • I think to finish off my thought from yesterday – the two images I posted work *well* in a magazine, a glitzy marketing presentation, an internet Infographic, or especially one of those internet political images that float around and people post on FB all the time that they picked up from Generally speaking I do not think the first chart works in a BI environment though – I’d still want a bit more that a quick-hit, quick-take away visual.

        When we as BI guys talk to customers, we should clearly be trying to determine what the essence the customer is trying to capture. based off of that, we can use the science of BI and visualizations to choose the appropriate one. Bar charts can show a variety of different things depending on how you arrange them as an example. This image here Chart Types is a solid guideline for the science part – probably should have posted this instead of the original image…

        Good conversation.

  1. Pingback: Last week’s OBIEE News | Addidici - 13 Years OBIEE and OBI Applications

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: