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.
I hate to be the contrarian here, but I actually disagree, I think this 'data visualization' is fantastic
Granted, 99% of the tea I drink is the tea I drink is Arizona green tea, and maybe I'm a little swayed by reading the 'Four Hour Chef' but I think this graphic has a lot going for it. I can absolutely see myself, in the store, comparing different tea varieties and using this chart to decide on the one I want.
As a consumer buying a premium product, the last thing you want is to spend a ton of money for a flavor/variety you don't like; I can totally see how this chart, clunky as it is, could help to remove doubt and make a consumer more confident about their first purchase and more educated/excited about subsequent ones.
I don't think the words are all that bad; granted they are a little out there, but no more so than some of the vocabulary associated with other premium drinks like wine, coffee, or beer. Plus, it reinforces the whole ingroup/outgroup dynamic that helps make premium brands so valuable
So, I have to admit - I can be a bit cynical about premium-brand marketing, even on the products I actually buy. I've been known to describe wine as having a "a strong taste of grape with a hint of stainless steel" :)
I think it's fair to say that the packaging reflects a premium brand, and that was 99% likely their goal with this visualization (it's really almost too small to actually read on the box). So, in that sense, it's effective.
I just don't think it conveys much information as is, especially without a comparison. In theory, you could create a five-dimensional flavor profile for tea, and it might even be useful, but without clear labels and some basis of comparison (a chart with multiple profiles), it's tough to see how that would communicate effectively to a consumer who was new to the brand/product.
One thing I didn't notice, mainly because this brand seems to change their packaging a lot, is whether similar teas in the series do have different profiles. I'll have to check this weekend.
Maybe it would help if they baselined it against plain water :))
And I think @Phantom is on the right track in this regard. As a return customer to the manufacturer, you could conceivably learn the profiles that jive best with your taste buds and, assuming there is at least some logic behind the chart, use the radar graph to quickly identify untried, but good teas (for your tastes).
So as a new buyer, using the graph would be mostly useless. But noticing it and mulling it over a fresh cup might cause that buyer to pull out the original box as comparison for the next purchase, and so on.
Then again, it could just be some guy's personal taste profiles turned into some box art to fill some empty space on the back.
I would be curious to see a follow-up post by Dr. Pete. I would trust a source that would attempt to give an unbiased analysis of the accuracy of the graphs.
An unbiased analysis would require transparency on how this chart was created, and the data behind it. Honestly, I doubt that exists, because the form factor and presentation suggests to me that it was little more than packaging decoration. Even as a return customer, I don't know what these dimensions mean or which of them I might want more or less of - there isn't any context or standard behind this. I bought this tea because it's a green/mint blend, which I can't translate into the profile at all.
I agree that a radar graph as a taste profile could work, in theory, but in this incarnation, it seems to be little more than eye candy. Even comparing 2-3 teas, I'm not sure I would gain any useful information.
To be fair, I will compare similar packaging and see if there's any rhyme or reason (my other products by this manufacturer have different packaging).
Phantom, I'm glad you wrote that.
As 'data visualization' goes, this choice of a radar chart to show ranking along a number of various scales is excellent. Now presumably, other flavors in that brand use similar graphs because they are more meaningful if used for product comparison. Bad data visualization would have been a 'bar chart' or 5 pie charts side by side.
The post does not discuss so much data visualization as it does graphic design elements - quite a different subjet. Even then, I would disagree. I find the whole thing quite clear, pleasant to look at, and it conveys that sense of a premium product. It doesn't mean I would buy that kind of product though.
In response to #2 - It can be seen as a major victory for Lipton that a company that produces $110/lb gourmet tea feels as if they need to include a measure of "briskness" in their copy. I worked in the specialty coffee industry years, primarily at small, independent, artisan cafes, and you wouldn't BELIEVE the pervasiveness of "bold" as a descriptor of coffee, due to the long-standing Starbucks ad campaign making it coffee's primary attribute.
Thing is, "bold" doesn't mean anything. At all. Customers would ask about our "boldest" coffee, and I'd have to ask, "Do you mean dark? Highly-caffeinated? DO you want a coffee that speaks truth to power?" Suddenly, we;re forced to describe our pricier, direct-traded coffee using terms introduced by a company that's entirely focused on mass production.
If customers are convinced that "brisk" is a thing, then brands have to either cater to it or debunk it, in my mind.
So, I'm learning a lot about the coffee/tea industry from this post :) I completely agree on the adjectives, though - it's amazing how, with enough time and money, you can make nonsense mean something to people. I guess that's effective marketing - I can't deny that - it's hardly authentic or grounded in real numbers, though.
It's also funny how you can play with a chart to communicate different meanings with the same data. I once had to give a brief on the status of a Marine Corps cryptolinguists, which was significantly below the required population. I remember stating before the brief "this chart is going to get a few pissed off looks" and then a Gunny took the top right corner of the chart (Powerpoint) and pulled it to the right. "Fixed, that ought to calm them down!" I won't say which version I used... :-)
Ah!
What a needed call to realism, Pete!
Every infographics or a graphics must be rightly communicative with minimalistic and creative design. Data Visualization is really important when communicating complex information in simple, creative and an easily understandable manner. A data visualization when goes wrong it can be really hard to digest or understand at times.
I added an update/apology - it appears that this diagram was more useful than I originally gave it credit for:
"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."
As you know Pete, I'm a big tea geek, too.
However, unlike you, Sencha is among my most favorite tea ever. Like uncovering some precious jewel, I was excited to learn what "umami" was back in my university's Japanese tea ceremony class.
My point here is that perhaps some fancy vocabulary is appropriate depending on the audience. This tea infographic tells me that the product is meant for someone who likes taking their tea beyond the usual fare. And if the vocab turns me off (whether I know what it means or not), then it's probably not my thing anyways.
You may have been suggesting that with your recent edits anyways.
just like 99% of infographics, 99% of visualised data on the web sucks ... good work pete
I hate infographics. Waste of internet space! Thanks for your comment, I was starting to think that I must be the only to find them pointless.
A great example that "GET TO THE POINT" marketing more that often works. In my opinion, a mediocre but perfectly communicated message or idea will always beat a poorly comunicated but beautiful message
"Get to the point" - marketing.. Great sentence, perhaps it could be Hop To The Point (HTTP) marketing just for fun...
Throw you design up on something like fivesecondtest.com. If the majority of your feed-back is "I don't get it," than your design doesn't make any sense (aka, it sucks). As a pseudo-designer, I know how hard it is to hear bad things about your work but I encourage you to accept criticism. If people can't understand your message, then your design is pointless anyway. :)
-Bryant
Why not do some Beta testing once the visualization is complete? Make some personas based on the target audience and take their feedback on how easily and what exactly do they understand by the visualization (Especially when you know the risks are high) !
At times, being SME's of the product / service, we sometimes forget what's most obvious to us may not be as obvious to a lay-man.
Every individual has a different way to look at things. You may not be able to satisfy all of them, but at least find out a common trend that would help in the Go / No Go decision.
KAS
Data visualization is certainly meant to convey otherwise dry and boring information in ways that are more pleasing to the eyes of audience. However, due to its easy-to-go-viral nature, infographics and other types of data visualization are sometimes overused. Some of them are too focused on design that that they often forget the main communication purpose. Great job to point it out, Pete!
The Taste Profile of your tea definitely needs more explanation or words as you mentioned above. If it wasn’t for your update, it would be hard for people who are new to the tea experience to understand the graphic. If it was to assist customers choose types of tea that complement their taste, it was not very helpful to unfamiliar buyers.
Trang Lam from Powered by Search
the main problem with so-called radar charts is that they are pseudoscience unless
a) the "categories" numeric or have intrinsic ordering
b) the space between the categories has meaning too.
For example, if there are (among others) 2 categories such as "taste" and "texture", the implied interpolation between taste and texture is utterly meaningless.
The next issue problem is that the impression of the chart can be radically altered by changing the ordering of the categories. Try this by reordering the categories of the image at the top of this posting.
Radar-charts: pseudoscience foisted off on an uncritical, scientifically, technologically illiterate public. Radar charts are no different from so-called charts that plot people's names (eg. sales numbers for a team) and interpolate between them (what does it mean to be 1/2 way between Mary and Pete? It's absurd)
Well, this is a good post. Nowadays is all about data visualization since data sets are growing and the new challenge is summarizing them in new specific actionable data.
I do like infographics, but often they have too much information in them and the text is so small you can't read it. Like content I often switch off half way through reading one.
Maybe there should be a best practice to keep them as short as possible.
lol. scooby is all mystery i love mystery so i love scooby doo. Sometimes too much watching movies can be applied in SEO. nice topic pete, keep it up. i love it.
:-)
I really like that diagram. Reallly like it. Taste Profile. Yum. I think that if I was a tea aficionado and saw that diagram many times for many different types of tea certain shapes would make me salivate or get dry mouth because I know what that tea will be like before I taste it.
Show it to a noob and he will not make much sense of it. However, he might assume that there is some sort of "science" behind it or that a team of gurus evaluated the tea on five different merits and created that Taste Profile chart as a holistic appraisal.
These noobs might remember in their schooling things like ternary diagrams from physical chemistry, color spheres from art class, beachball diagrams from seismology, and n-dimensional space from statistics. They were, at that time, not willing to think hard about the subject so they simply blew off that part of the lecture as soon as it got a little bit away from them.
So, Joe Schmoe is likely to see that diagram and think that he is drinking nectar, with a taste profile that has been evaluated and approved by spiritual gurus and the placebo effect will kick in.
My reason for liking the diagram is imagining how it can be used to describe a five-variable system in a graphical presentation which someone familiar with things can - at a glance - immediately know what they have.
Thanks!
It has become way too easy to make graphics, and now it is like text with pictures.
Our industry gets all excited when things get easier, but we forget that that also means the competitive advantage goes away. When everyone can just spit out an infographic or visualization, then we all have to work a bit harder to stand out.
Absolutely. And graphics should support the information, so that together they communicate something more.
Hey Pete! I found this to be a great read for my morning. Our office is split with coffee and tea (Brooke is the coffee drinker :)) so this post resonated with me. It also got me thinking of a tweet you did a while back with a McDonalds sign that said something like "buy our crap".
I guess why that popped in my head is that the intent is all the same. We want your business. It's just that since people can be duped by big and shiny, sometimes it's (unethically) easy to leave out some important facts -- like our "French" tea is made in Wisconsin. #GoVikings Thanks again,
What a great post and a much needed call back towards the merging of data insights and design. I am a huge fan of data visualization because it brings fantastic insights as well as being easily marketable (both internally and externally). However, like most content, the insights and value it brings should be paramount to an aesthetic, since the purpose isn't to post it on your fridge or hang it above your couch. If it's not doing anything for the end 'user' then whats the point?
Perhaps the lesson here is to be intentional about your visualizations. :-)
It's hard to dig into much useful here without context of the packaging, etc.
While the diagram is aesthetically pleasing, agreed with Dr. Pete that a Wild-West-esque take on "scales" and "scores" is risky business.
Great post Pete. Saw your twitter discussion on it and hoping that coffee people dont read your work :) Some people work best in visuals, but the visuals need to convey the message effectively, which is certainly not the case in the above scenario. I love strong visuals, particularly good infographics but the definition of infographic is getting more blurry with 'infographic light' being what is produced. Great content , whether in content form or infographic is what is going to give the best value to you and your users.
Great little post on bad data visualization!
Infographics are also a place to start. I often see poorly made infographics, that tries to deliver stunning visual data representation, but is clearly more confusing than helpful.
Thumbs up for useful data visualizations!