How Not to Visualize Your Data
Posted by Dr. Pete
Lately, I’ve been seeing data visualizations everywhere, including the products in my own kitchen. This week, I had sightings on my tea and my tortilla chips. This is a story about the box my tea came in (for the sake of my marriage, I can’t disassemble the tortilla chip bag until it's empty), and how sometimes we take marketing too far. Over the weekend, I discovered this "Taste Profile" (the top version is a recreation, since the real graph was only about 1” tall, but all details are accurate to the original):
I’m not attacking the company that made this, and I’m not going to “out” them here – their product is actually pretty great. I just want to use this visualization to illustrate some of the wrong ways to do things, in hopes that we can all raise our game a bit.
But It’s So Pretty!
I admit – the earth tones are nice, and it’s not entirely unappealing. I guess, for a moment, it made me feel better about shelling out $11 for an ounce-and-a-half of leaves. Maybe that’s even good marketing, although I really doubt this 1” tall graphic on the back of the box has ever swayed anyone’s decision. I’m not trying to say that it’s an ugly picture. The problem is that it’s a pleasant distraction disguised as meaningful data.
The job of a data-visualization is to communicate an idea better than the raw data itself could. Of course, that also implies that there’s actual data behind the visualization. So, how do we get it wrong?
(1) Pick the Shiniest Style
We all know that the best chart style can be summed up with two words: “big and shiny!” The radar chart above is pretty shiny – it’s like I’ve discovered some lost continent of tea with my smooth jazz submarine. The problem is that, ultimately, I don’t know what that shape means, and I don’t have anything to compare it to. A radar chart is at its best when comparing two or more profiles. Pick the right tool for the job, not the one that looks the most impressive on your utility belt. Batman is a friend of mine, and you, sir, are no Batman (disclaimer: I don’t know Batman).
(2) Use a Lot of Fancy Words
Umami is the exotic fifth taste (beyond the classic four of sweet, sour, bitter, and salty) – it’s a Japanese word meaning “Haha, I can’t believe I got you to eat sea urchin!” To be fair, at least it has something to do with taste. I honestly have no idea how “Brightness” or “Briskness” apply to tea, and if they do, what the difference is between the two.
I do know that Lipton has spent a lot of money making us think their tea is brisk, which raises another point – why do you want to compare your $110/lb. gourmet tea to Lipton? Even “Aroma” is a bit ambiguous – do I want a lot of aroma? What if it’s the aroma of some bad umami that I forgot to put in the fridge last night?
The goal of a visualization is to simplify information that’s too complex. If you have to make up big words to do that, then you’re missing the point.
(3) More Words? Yes, Please!
What really brings a visualization together is to explain each of your terms with even more words, preferably ones that make even less sense. Now, please understand – I have no issue with the French. I think Paris is lovely, it’s cool that you helped us win the American Revolution, and I’ve never eaten “freedom fries”. This product wasn’t made in France, though, and I didn’t buy it in Quebec. The company is based in Milwaukee, Wisconsin.
Translating every label on the graph into French isn’t just meaningless – it’s pretentious. These secondary labels only serve to add visual noise and make it harder to pair the main labels to their data points.
(4) Keep the Mystery Alive
Everyone loves a mystery – you don’t hate Scooby Doo, do you? If you can make your product mysterious enough, everyone will think they need it. Sadly, sometimes smoke and mirrors is all a product has to offer, but in this case the product is really quite good. Adding pseudoscience to the label doesn’t create intrigue – it just makes me wonder if the marketing team is drinking their product or smoking it.
Communicate, Communicate, Communicate
To be fair, this 1” graph was little more than a decoration on a box, and it does that job perfectly well. Unfortunately, I’ve seen similar graphs (and worse) in blog posts, research papers, and even reputable newspapers. Every day, it gets easier to make sexy charts, illustrations, and infographics. It’s ok to create something beautiful, but we have to remember that our first job is to communicate. A data visualization should convey useful ideas quickly, because ultimately that’s our job as online marketers. So, think before you open Photoshop.
Addendum: So, I've learned that "cupping scores" are not uncommon in the gourmet coffee industry. Here's a 10-factor radar graph (hat tip to @jimbeetle). I just have a hard time seeing this as anything but a way to justify premium prices with pseudoscience.
Update & Apology
I don't usually add a section to a blog post after the fact, but in the interest of fairness and TAGFEE, I checked other green teas by this manufacturer and found that they did, in fact, have their own unique profiles. So, I looked at some teas that I historically don't like (such as Japanese Senchas), and found they were high on the "umami" scale. That was actually pretty interesting. Then, I picked a tea with a similar profile to the one I like – high in "brightness" and low in "umami" (although, in this case, more "brisk"), and I took it home. It was, in fact, delicious.
So, while I never intended this post to pick on the manufacturer (their tea is excellent), I do apologize for suggesting that the diagram was little more than packaging decoration. I still believe the takeaways here are useful, and I don't think a casual consumer could interpret this visualization, but there does appear to at least be some method behind the madness.
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don’t have time to hunt down but want to read!