1. Serving the Presentation Without the Data

Which comes first: the presentation or the data? Oftentimes, in an effort to make a visualization more “interesting” or “cool”, designers will allow the presentation layer of a visualization to become more important than the data itself. The visualization captured below is an unfortunate casualty. A considerable amount of work went into it and there are parts that are informative, like the summary counters at the top left. However, without a scale or axis, the time series on the bottom right is meaningless and the 3D chart in the center is even more opaque. Tooltips (pop ups) would help, if they were there. Instead, this looks amazing, but does little.

(source)

2. Showing Too Much Detail

We all know the feeling of finding a dataset that is rich and easy to visualize, with numerous usable categorical and numerical fields. The temptation is to show everything at once, and allow users to drill down to the finest level of detail. Often, that actually makes a visualization superfluous because the user could simply look at the dataset itself if they wanted to see the finest level of detail. The trick, then, is to show enough detail to tell a story, but not so much that that story is convoluted and hidden. More data can be revealed as the reader progresses through the story. This visualization could have been great but there is so much detail it is hard to garner much information from it.

Browse more infographics.

(source)

3. Not Explaining the Interactivity

Enabling users to use and interact with a visualization makes it more engaging and engrossing. However, without telling them how to use that interactivity you risk limiting them to the initial view. How you label the interactivity is just as important as doing it in the first place. Usually, informing the user at the top of the visualization (or the part they will see first) is good practice, as is calling out the interaction on or near the tools that utilize it. This visualization is actually very interesting, but without labeling the interactivity, it is easy to overlook the fact that you can click the words at the bottom of the screen to change the view. Even using a common design concept such as underlining the words to associate a hyperlink, would have been helpful.

(source)

4. Failing to Experiment

Often, your first idea is not the best idea. It is easy to get excited about a visualization and then stick to the first vision that came to your mind when you saw the data. However, it is always best to start a visualization with a blank slate of ideas. Then shift perspectives: try ten or twenty different configurations and types of chart before you settle on one. The best part about experimentation is that it often forces new findings out of the data. This chart is not ineffective, but it would be much better if it used bar or even area charts to display its information. It is actually difficult to compare the magnitude of the different parts of the “blob” with this type of view.

(source)

There is no perfect visualization, but if you can manage to stay away from these four mistakes, yours will have a much better chance of getting close to perfection.

Ross Perez is a Data Analyst with Tableau Public, a free tool which allows people to put their data on the web in interactive charts and graphs. You can connect with him on Twitter.

I shall not use visualization to intentionally hide or confuse the truth which it is intended to portray. I will respect the great power visualization has in garnering wisdom and misleading the uninformed. I accept this responsibility willfully and without reservation, and promise to defend this oath against all enemies, both domestic and foreign.

It is doubtful many people use visualization to intentionally mislead, but there are countless examples of visualizations and information design going wrong, anyway. This oath seems to carry the underlying message that one must create visualizations that not only use accurate data and avoid 3D pie charts, but are purposefully designed to tell a clear story and convey information effectively. In other words, any old graph will not do — every dataset deserves its own canvas.

The easiest way to explain this is to look at a well-made visualization and describe why it is effective. Obviously, no viz is perfect, so take this as a “let’s point out what works and what doesn’t” discussion, rather than a dissection of the viz archetype.

This one was originally published on Geekwire and shows the (rather disappointing) collective performance of last year’s tech IPOs.

The visualization tells a story: 2011 tech IPOs are tanking. The moment anyone looks at it, the story is clear, and there are a couple of reasons for that.

• There is a clear title;
• The annotation clearly tells the story “Start Strong and Then Fall”;
• You can see the story played out in the line chart itself;
• To confirm the finding, you can examine the individual results for each company right underneath the line chart;
• The color legend is clear and also displayed (a common problem with many graphics).

This piece is particularly interesting because it tells a very simple story, yet the data itself is complex. Imagine the myriad ways that one could show the aggregated percent change for twenty different companies. The author of this visualization experimented with different views and arrived on the two that told the story most completely, most effortlessly.

And they did not stop at just completing the visualization. They labeled the interactivity “Include These Companies:” and explained the complex pieces in the visualization (the slider that filters companies based on trading days). Perhaps most importantly, they told us how they got the data and when they last updated it, so should we want to, we can go check it for ourselves.

Great visualization is not simply a chart that shows the data; it is a chart that tells an accurate and interesting story. Creating great visualizations is not easy, but by examining well-made examples, it becomes a little easier.

If you would like to learn more, Tableau has compiled a series of videos on the Tableau Public site that go into significantly more detail.

Ross Perez is a data analyst at Tableau Software. You can connect with him on Twitter.