skip to Main Content

“The customer is always right”

-Anyone who knows their business

A business intelligence feud!

A few weeks back, a BI analyst posted a list of factors to consider when selecting a new BI tool.  Shortly thereafter, a BI blogger posted a scathing critique of that article.  The analyst, for his part, then blasted the blogger on Twitter (and curiously, re-linked the blogger’s offending post).  By BI community standards, this is a feud of uncommon proportions.

I don’t really want to get in the middle of such an entertaining exchange, but I feel compelled to offer at least a short comment.

You see, the analyst had the audacity to suggest that the number of visualization types (charts, gauges, etc.) supported by a BI tool is an important factor for potential customers to consider.  Perhaps he was unaware of the newly prevailing wisdom:  fancy visualizations are bad for you.  Simpler is better, 3d is a distraction, pie charts are passé,  and gauges/stoplights are toys.

I think that point of view is at least partially misguided, and while rooted in scientific fact, it misses something fundamental about human psychology.

The prevailing wisdom of the chart police – three examples

1) Use Bar Charts Instead of Pie Charts

Three Color Bar Chart Pie Chart

Simplest Visual Comparison

Same Numbers, Harder to Compare

There is research proving humans are good at judging the relative lengths of objects, but bad at comparing the relative sizes of areas.  (Although I’d submit that, in college, my friends and I were masterful estimators of pizza subdivision, so perhaps the human race as a whole is just out of practice).

2) Avoid 3d and Other Effects

…because again, they just make visual comparisons tougher.

3d Bar Chart Pretty Bar Chart

Not Acceptable

Probably Earns You a Lecture

  Basic Bar Chart  
 

Boring Yes, But Efficient

 

3) Gauges and Stoplights are Just Plain Silly

KpiGauge200Up  Stoplights Modified

…and apparently animated bubble charts have also come under fire lately.

Is Your World Frictionless?  Mine either.

Remember in physics class, how we were often instructed to ignore the effects of friction and/or air resistance?  That was helpful, in an instructive way – we learned useful principles.  But you can’t apply those in the real world without factoring friction back in.

A similar thing is happening here.  And the friction that’s being ignored in this case is the human attention span.

Before someone can begin breaking down and interpreting a chart or other visualization, they must first decide that it’s something worth paying attention to.

powerpivot icon And most people love eye candy.  They just do.  It’s fun to look at cool pictures, and fun draws people in.  For example, some of my friends worried about the 3d chart appearance of the icon I chose to use for my site.  “It runs counter to the prevailing advice about charts” was the concern.  (You might notice that the PowerPivot field list never creates 3d charts – discretion is the better part of valor, even in software).

But when I took a vote amongst my friends and colleagues, this icon was the overwhelming favorite, with more than 2/3 of the respondents selecting it as their first choice.  It was not well-liked by the other 1/3, however, who happened to be the people I already regarded as the pickiest people I know 🙂

Show Some Restraint, but Always Play to your Audience

Here’s what I now believe:  the Chart Police are loud and influential, and they have good information to share, but they do NOT represent a broad cross-section of the public.

Wisdom dictates that you give your audience what they are most likely to engage with.  So, if your report audience is mostly normal people, feel free to spice things up a bit.  Don’t overdo it to the point that the original point is obscured, but make it fun.  Make it attractive.  Because your message won’t mean a thing if they don’t engage with it first.

If you know your VP likes their reports to contain pie charts, or 3d charts, or even animated bubble charts, by all means, deliver them.  Don’t feel like you have to re-educate everyone about the sins of these things.

Of course, if your audience a bunch of college professors or professional graphic designers, I recommend following the chart police guidelines closely, because they very well might think you an amateur if you do not.

This isn’t a question of right or wrong, really.  You just want your message to have the highest uptake possible.

Personally, I’d love to see studies that measured both the accuracy of the information conveyed by various visualizations, combined with the level of audience engagement.  I really do think the Chart Police are onto something, but I think the story is unfinished as it stands.

Further Reading

I generally think of Edward Tufte when I think of the chart police.  I’m not at all certain that he is the source of the venom that I see on the topic, though – I tend to think that he just identified some things that had previously been ignored, and then people who read about that picked up torches and pitchforks.  Regardless, he does fantastic work, and notably is also an artist.  Browsing his site is definitely a good use of time.

I also thoroughly enjoyed Eye Candy is a Critical Business Requirement.

Back Next Week

I’m visiting relatives in Chicago right now and don’t have access to my PowerPivot machines.  So this was a good opportunity to cover this chart topic that’s been on back burner for weeks.  Next week though, I’m moving ahead with DAX, might be finishing (!) the Great Football Project, and I have a couple of other reference topics about PowerPivot that I’d like to start compiling.

See you then!

Rob Collie

One of the original engineering leaders behind Power BI and Power Pivot during his 14-year career at Microsoft, Rob Collie founded a consulting company in 2013 that is 100% devoted to “the new way forward” made possible by Power BI and its related technologies. Since 2013, PowerPivotPro has rapidly grown to become the leading firm in the industry, pioneering an agile, results-first methodology never before seen in the Business Intelligence space. A sought-after public speaker and author of the #1-selling Power BI book, Rob and his team would like to help you revolutionize your business and your career.

This Post Has 18 Comments
  1. Disagree with 1) and 3). Agree with 2)
    I may change my mind later, but I believe you need to pick the correct chart that’s appropriate for the things that you need to report.

    There are some reports that I like to see pie chart but not a bar chart (e.g my spending habits by month)

    A gauge may be fine in some cases (yeah it could be a less fancy gauge) but if I’m tracking certain measurements and want to see red zones and it’s relevant to the report (i.e. Not on budget tracking , but maybe on some real time throughput reporting) It could work.

    Traffic lights are a waste of time – better off using simple background colors on cells.

    1. Hi Leah 🙂

      Just remember that a) rules 1-3 are not mine, they come from the Chart Police and b) I’m in favor of anything that gets your audience’s attention AND successfully conveys the information. Even if it’s a stoplight 🙂

      -rob

  2. It’s true that there is research showing people understand simple, and particularly linear, representations better than more complex ones. I did some of that research myself. There are times when simple linear representations aren’t so practical, though. When you need to compare values that span a wide range – such as amounts from a few dollars to thousands or millions – a bar chart would have to be very large for both the large and small amounts to be visible, perhaps too large to fit an ordinary page. There is also the issue of keeping the attention of the viewer. If the audience likes pretty graphs, they like pretty graphs. The analyst has to deal with those expectations. So while I wouldn’t say that I judge a tool by the number of graph types, I certainly value having a variety of types available, the ability to combine different elements as needed, and the option to “pretty up” a presentation to suit the situation.

    1. Thanks for the insightful comments, Meta 🙂 I find this whole topic quite fascinating and enjoy hearing the opinions of experienced pros such as yourself. I also just read your “5 myths” article on your site and loved it.

      1. summitcloud,

        That approach gets the numbers into the chart, but it does so at a cost. The chart is no longer linear, and so the perceptual benefits of the simple linear representation are lost.

  3. I feel your pain, I may also have been one of the chart police at some point…

    Ultimately you have to reach a compromise you can live with. Eye candy may be required but it’s up to you to make sure that it doesn’t mislead or downright lie.

    Really we need to educate our colleagues to recognise misleading graphs and demand better.

  4. I understand your point but I’ve found that following the advice of the so called chart police (including their advice on table and document design) you end up with very attractive and clear reports.

    The managers I’ve worked with often find these reports to be a breath of fresh air compared to what they’re used to. I haven’t had a request to add 3D, gauges or traffic lights yet but maybe I’ve just been lucky. There once was a request to add smilies though but thankfully nobody else thought that was a good idea 🙂

    1. Makes sense to me. I think it’s more a reaction on principle for me. I’d like to see more acknowledgement of “get your audience’s attention” and therefore some guides on “here’s how to grab eyeballs without sacrificing clarity.”

      For several years now I have done most of my MS-internal PowerPoint presentations with the blank, black-on-white slide theme. Similar to what you say above, I found my audiences appreciated that cleanliness after years of exposure to heavily stylized slide decks.

      But when I go to a conference or the like, well, I know the audience expects slicker – the blank theme wouldn’t jive with the formality of a planned event.

      thanks for the comment 🙂
      -rob

  5. Patrick Husting has been doing some posts lately on visualizations that are relevant to this chat. In the 3rd post, notice how Excel Services helpfully trims 3d effects for you when publishing to the server 🙂

    http://bimvp.com/blogs/bsm/archive/2009/12/10/great-looking-pie-chart-the-american-diet-34-gigabytes-a-day.aspx

    http://bimvp.com/blogs/bsm/archive/2009/12/15/i-really-love-good-displays-of-analytics.aspx

    http://bimvp.com/blogs/bsm/archive/2009/12/16/search-engine-market-share-in-excel-14-and-excel-services.aspx

  6. There is one major problem with the ‘grab their attention’ approach to data visualisation, that is that the lazy recipient will see the obvious and miss the real message of the graphic. IMO, eye-catching visual effects pander to the lazy recipient, and it is far better to make the information presented clear, and if that means that the recipient has to work a bit to absorb the information and that then means that the lazy recipient is incapable of getting the information, so be it. Surely it is better to not get it than to get the wrong information? Not everything can be absorbed in 2 second bytes.

    As to your point about grabbing your audience’s attention, very relevant, but are dodgy visualisations the way to do that? I would have thought that you should; get the right audience; give a clear opening about what you are presenting and why it is relevant/important at that time; and keep the presentation brisk, relevant, and focussed.

    1. I tend to think that most people are lazier than we’d like 🙂 OK, let’s call them “overloaded and busy.” As a designer of software over the past decade, I will say that the cases where I built software that made the user work… were the least successful of my designs.

      How about we just make sure our reports and presentations are fun and inviting in some way. I don’t endorse the really silly stuff. But even that depends. Ever attend a Donald Farmer presentation? He keeps people laughing throughout, while never sacrificing message. As a result, the audience is 100% engaged and absorbing everything.

      There are of course limits and context to apply. I suspect that you and I would agree on most specific examples – “yes on this, no on that.” I just want to make sure that the cult of “make it completely dry and robotic” doesn’t go completely unchallenged.

      I am pleased with the direction of this discussion. Keep the opinions coming 🙂

  7. We don’t get to pick the audience. They are our bosses, our customers, our constituents. If we hope to influence them, it’s up to us, the presenters. Adapt to suit your audience or suffer the consequences.

    The presence or absence of shading on a bar chart, the use of a bar chart instead of a pie, colored backgrounds versus white, none of these details ever made or broke the success of a presentation to influence an audience. The real success comes with having something worthwhile to say, focusing on the most important material, relating the message to the concerns of the audience, organizing the presentation clearly, being prepared to address questions.

  8. I think that it’s important to separate charts that are useful for BI and charts that may used for eye-catching presentations. Most arguments fail to take the differences into account, but the distinctions are critical. From the BI perspective, any “ink” not directly related to the data you’re analyzing is distracting and simply gets in the way of analysis. We engage in BI for survival and to attain a competitive advantage – not to impress our bosses or an audience. One key tenet in analysis is to use the best chart (or combination of charts) pertinent to the analysis. This sounds obvious, but is rarely applied in practice.

    For BI, I have never found a scenario where a pie chart represented data better than a bar chart. If you want to use a bar chart and the data has a wide range, use a log scale. Instead of gauges, use bullet graphs. 3D charts are both hard to read and (depending on the chart) mask some of the data points (or worse, distorts the data e.g. 3D pies). Basic line charts are the way to go for analyzing trends over time. If you have to analyze a lot of data, you need a chart that highlights anomalies or outliers – individual data points are of little consequence with a large data set.

    So how does the number one BI tool (Excel) fare when it comes to visual analysis? Excel charts haven’t changed much in over 20 years, demonstrating how little regard has been paid to visual analysis. In 1997, we got 3D cone, cylinder & pyramid charts. In addition, we got radar charts, bubble charts and stock charts. In 2007 we got eye candy. In summary, in over 20 years, Excel has added no meaningful enhancements in charts for BI.

    To its credit, Excel’s charting tool is quite flexible and you can coax almost any chart out of it. Alas, most of the techniques are known only to experts. For the other 99.99% of users (i.e. the folks who the product is designed for), these techniques remain undiscoverable. Although the Ribbon was created to make features more discoverable, it cannot help in this regard.

  9. As an aspiring chart police cadet and member of the ‘Picky’ 1/3 I’d like to add my 2 cents 🙂

    Most of the points I would make have already been made in the great discussion in the comments but here is what I’d like to add.

    In the Stephen P. Anderson presentation you link to he talks about the importance of aesthetics, that being great visual design that supports the idea that you are trying to communicate. “Eye candy” is something else; it is the more refined relative of “lipstick on a pig”.

    When I think about charts I often think about one of their highest profile uses for business which is in annual reports. Here is an example from Goldman Sachs 2008 annual report. [PDF, 1.1MB]

    I always try to keep in mind this quote by Antoine de Saint-Exupéry, “You know you’ve achieved perfection in design, not when you have nothing more to add, but when you have nothing more to take away.” After all, I, like most designers, are just eye-candy junkies desperately trying to stay on the wagon.

    Thanks for the post Rob.

Leave a Comment or Question