Tom Chi  

Playing the TUBA

October 17th, 2003 by Tom Chi :: see related comic

Beauty. Usability. The discussion goes back a long way. I mean a REALLY long way.

Usability has been around since we started making tools… at least a million years and likely longer since chimps and even some birds use them. And as for beauty, we’re looking at more like 100 million years as finding a mate involves some sense of “beauty” (defined as the happy end of any scale of aesthetic evaluation).

So we have these two very old concepts coming together because we gotta sell things for cash (a relatively new concept), and the upshot of this is that we have at least 3 things to optimize for: usability (U), beauty (B), and revenue. Optimizing for revenue could also be called marketing/advertising (A). Oh, and for completeness, we’ll need to add one more axis: technical innovation (T). Now just rate each of these attributes from 0-9, toss in a fondness for brass instruments, and we’ll call the whole thing the TUBA score.

Tech : Usability : Beauty : Advertising

Let’s see what we can do with this.We’re going to take these axes, and use them to break down some sucessful products to see what it takes to suceed. For example, Pepsi might get a TUBA score like: 2:7:3:9, while Apache Web Server might be 8:6:2:0. Notice how Pepsi is pretty poor at technical innovation and Apache is not great at advertising, yet both are sucessful. Thus, it’s clear that you don’t need to win in every category to have a good product, and to this regard usability and beauty are not at odds in any fundamental way. In order words, they are not mutually exclusive and it is possible to do fine with mostly one or the other. Sometimes you even get both! (e.g. an iPod might be 8:7:6:9).

So why do usability and beauty fight? Well, first off, in any real world project there are limitations on time and resources. You only have so much time to improve things in your respective category, and thus if your product cycle is dominated by tech and/or adverstising, you’ll see usability and beauty getting squeezed. In addition to this, there can be *some* intercategory dependences. For example, if technical innovation is high, it’s important to keep usability up there too (you don’t want to baffle user with too much tech). Conversely, if technical innovation is low, you have more leeway to focus on beauty (designing a new fork or a new chair, for example). Other examples: If your advertising is amazing, sometimes you don’t really need to be so technically innovative, and other times if your product is incredible you hardly need advertising.

While it is nice to strike on all pistons, it is not required to create a successful product. That’s why it is totally sensible for Norman to say that aesthetics can trump usability. The reverse is true as well, and more generally, any high score on any TUBA variable has the potential to carry a product to success: tech can trump the other 3, advertising can trump the other 3, etc.

The caveat to all this is that usability and beauty both have a minimum level to be achieved. If something is absolutely unusable, or absolutely ugly, then it just won’t fly. Beyond that, the level to which you must invest in each space will be dictated by the problem you are solving. In the technology world, usability will always a concern (since new concepts need more usability to be understable), but with teapots (which are imminently familiar and *not* new), you can be almost completely aesthetic and Norman will still buy it. :p

Still, anyone who works in any these disciplines dreams of that elusive confluence where all come together to form an incredible product. Afterall, it is the incredible products which capture our imaginations and redefine how we approach the world. Practically though, most of us need to pick our battles, as it is better to excel at *something* than to be mediocre at everything. I don’t think that Apache web servers would become much more sucessful with an incredibly beautiful interface, and Pepsi won’t win too many more customers by making their cans 5% easier to open. Alas, we can still dream. Or we can try to work at IDEO.

15 Responses to “Playing the TUBA”
Meri wrote:

I love this article. I think that it has really given voice to a concept I have had a hard time articulating for years! Thank you and can I use the TUBA-speak please?

Tom Chi wrote:

Sure, definitely go ahead and use it. Don’t forget to tell those people to read the comic though!

Meri wrote:

No problem — I’m thinking about this whole thing and will direct back to you on my blog when I get the post in my head out and translated into electrons ;-) If I do a presentation can I use the comic, so long as you’re referenced as the owners? (Does it show I know NOTHING about Copyright? ;-) )
–Meri

Steve wrote:

Agree heartily with all your points but one…

I think — no, I’m sure — Pepsi would gain sales by making their cans 5% easier to open. And, with the volume that they do, even an extremely small improvement can turn into an extremely large number of $$.

Real World Example: I recently read of a mini-boom in sales of soda cans due to the change from bulky boxes of canned soda to the new “fridge paks” that take up much less space and dispense soda one can at a time. I know I love ‘em!
Steve

Kevin Cheng wrote:

Meri, the best thing to do is e-mail us for permission and we’ll happily give it. I’d give it to you here and now but then, anyone can leave a comment and say they’re KC or Tom =)

Tom Chi wrote:

Fridge paks is a nice example but it should be noted that it involves a change to the form factor of the packaging, which affects beauty and also gives advertising something to hawk.

To truly isolate a “5% improvement” in usability, imagine that pepsi makes cans slightly easier to open, but does not advertise the change nor does it redesign the visuals to indicate the change. I bet they would see only marginal improvement to sales. For the effort, they’ll get more mileage out of a hip new visual design or by launching a new marketing campaign where they give out iTunes songs, etc.

Frank Spada wrote:

Great article, Tom. I tend to look at advertising a bit differently. I see advertising as everything that a company does directly and indirectly in giving the product visibility in the marketplace. Apache, for example, actually got a lot of advertising, in the form of word of mouth through an ideal target user base. The most extreme form of non-advertised products I can think of are generic supermarket products. But even those have a stealth advertising strategy of piggy-backing on their brand-name counterparts - plus being on display at supermarkets for the bargain hunting consumer. If a product had truly no advertising it couldn’t exist.
I like your point about specializing in one of the TUBA criteria. Classic marketing principles mirror this idea - you either differentiate yourself from your competitors by boosting the value, quality, or customer service of your products. In the end, it’s all about finding the product that’s just “good enough” to sell and edge out your competitors.
Lastly, I spotted an interesting 5% improvement in pepsi cans at a street vendor in Italy recently. A very simple idea - a thin shrink-wrap piece of plastic covered the top of each can so that you could remove it and drink from a clean can! The improvement wasn’t directly advertised or visible; but I would certainly buy it again.

Don Norman wrote:

TUBA is clever and makes logical sense. Alas, I suspect it is wrong.

This is very similar to an approach that I have tested (the test is still going on.) Basically, it isn’t enough to formulate a logical analysis of people’s behavior: the formulation has to be tested with real people making real decisions about real products.

TUBA assumes several things. One is that the four dimensions – T, U, B, and A – are independent (orthogonal). Second is that the scores on the different dimensions combine in some known manner. Third, that the evaluation of any product along the four dimensions is consistent for all people. And fourth, that the decision rule for preference of one product over another is consistent across people (or even consistent for one person over time). My studies indicate that none of these are true. Too bad — I was hoping it would work.

The dimensions do not appear to be independent. We find that decisions on one dimension affect decisions on another.

It isn’t at all clear how the scores on the different dimensions should be combined: additive? No. City block or Minkowski metric? No. theory? Decision by elimination? (which is non-linear).? Maybe.

Worse, we have found huge inconsistencies across people for the same object and within people over time. And the stated preferences often conflict with the final choices.

Indeed, if there were consistencies, we would not expect to find so many different brands all doing the same thing on the marketplace. Why so many brands and models of automobiles? VCRs? Coffeemakers? Water? If the decision criteria were as simple as TUBA, then a few would dominate the rest (because their TUBA scores would fit people’s needs nicely). In fact, people buy one item today and a very different one tomorrow.

Take the example of the Apache web server. When there is essentially no competition, functionality and cost trump everything else. Beauty? For a web server? So what? But suppose there are 50 brands of webservers competitively priced and featured. Then beauty might very well matter. Suppose webservers are commodities, where everyone runs one on their refrigerator or TV set (or more likely, on both). Beauty might matter.

So the requirements for a product vary with time. VisiCalc, the first spreadsheet, was ugly and difficult to use. It was an instant success. But when Lotus 1-2-3 came along, VisiCalc was dead. (The actual argument is a lot more complex and political, but it is also the case that the deficiencies in VisiCalc could no longer be overlooked once a superior product arrived).

So, I really applaud the effort. It is on the correct track, But being logical is not enough in this business. One must also validate with empirical research.

Don Norman
Nielsen Norman Group

MeriBlog wrote:

The Importance of Training … and Why It Is Not Enough

Training is a fantastic idea. In order to secure real adoption from any workforce, or actual realisation of software’s worth, it is an absolute necessity. But it is not the be-all and end-all and it is definitely NOT a replacement…

Meri wrote:

Although I can see that perhaps this is not the be-all and end-all of design, it seems to be that it would be a very very useful framework to sell the initial ideas for products, or more likely to convince the teams alreading developing a product to pay attention to whichever of the TUBA elements they are not currently focusing on.

For instance, I have had a massive uphill battle (as most HCI advocates tend to) convincing my business-oriented IT team to consider the usability of their applications at an early stage. Since we primarily develop bespoke software (or modify existing software for our organisation) their focus is traditionally on “Let’s make sure it works and then train the organisation to use it properly”.

I’ve chosen to keep the rest of my training rant to my own blog. My point, though, is that although the TUBA framework might not be everything, it is definitely a starting point in conversation and a lot more elegant and eloquent a weapon than most of the others in my “Usability is a good thing” toolbox.

Matthew Oliphant wrote:

[assumptions aplenty]

Is it just me (answer is yes) or is there some inherit need to burn away subjective documentation and turn it into “objective” numbers? My company is doing something like this, and I am not sure if it is a good thing. I realize management, in all it’s many forms, is probably what drives this. “Just give me the numbers.” The mantra of those who wish not to be involved.

I like the concept behind this nascent idea. I think, as Don said, the dimensions are not independent. And actually I didn’t take them to be on first reading. I also think there are probably 7 more dimensions that could be applied (which would add a level of synchronicity to this month’s SciAm cover story).

If nothing else, this does give a way to show the categories of study that must work together in order to derive a “good” product/service. Even if the company never focuses on beauty, the idea (my thinking about the idea) behind this is that they take it into consideration and document why they chose not to purse it as a focus.

I also see this as a way to organize a competitive analysis. Company A is a 4:3:7:7. Company B may decide it can improve it’s part of the market share by focusing on surpassing company A’s 4:3, but not try to outdo the 7:7. Eh, just a thought.

Lastly, logic outside the circle of HCI-minded people doesn’t usually work and I think this has something to do with, “Just give me the numbers.” It’s probably also why companies tend to segregate roles so people can become “experts” in their area. The part that is usually missed is bringing these roles back together again. Perhaps TUBA, or it’s coming iterations can supply a framework for that to start happening as it is very needed by most companies.

Tom Chi wrote:

To respond more directly to Norman’s points:

1. TUBA assumes dimensions are independent.
2. Scores on different dimensions must combine in some known manner.
3. The evaluation of any product along the four dimensions is consistent for all people.
4. The decision rule for preference is consistent across people (or even consistent for one person over time).

1 - I stated that there would be practical dependencies between the axes. The axes are definitely not independent, but they are certainly different enough to be evaluated separately. I could run a usability test on time on task, and the result could be intepreted without having to think about beauty. So even if there are somewhat dependent, they are testably independent.

2 - I agree that there is no way to combine these four numbers into one supernumber. But I would tend to think that such a supernumber would be meaningless or at least hard to interpret. It reminds me of the statistics that rate which countries are best at math and science education. All those studies produce is a ranked list, but what does the rank mean? How was this evaluated? Clearly all the real data is hidden behind the evaluation function which created the final score — so I tend to see the final score as not as important. It you want to base your strategy on a study, you would definitely want it to produce more than a number.

3 - It’s ok that different people will evaluate these numbers differently. You could remedy that problem by testing a number of people in your target market, and running the stats to see where opinion lies.

4 - On this point, I agree. TUBA does not take into account changes due to time, or special conditions afforded by products being new (or old). Time would be a valuable addition to the model, and Norman provides some excellent points as to how time can change the dynamics of the axes. I discuss this a little more in my subsequent post.

Roland Tanglao's Weblog wrote:

Playing the TUBA - Assessing beauty and usability

(SOURCE: OK/Cancel: October 17, 2003 - October 23, 2003 Archives )- Don’t know if TUBA has the chops :-) but there are certainly some good concepts in this discussion.

Don Norman wrote:

Tom says:

“I could run a usability test on time on task, and the result could be intepreted without having to think about beauty. So even if there are somewhat dependent, they are testably independent.”

Are you sure? This is a debatable point — there is evidence both ways. But there is some evidence that if people find something attractive, they find it easier to use. As I said, this is controversial at the moment, but it points out that this is a difficult area — we have a lot to learn.

See: Tractinsky, N., Adi, S.-K., & Ikar, D. (2000). What is Beautiful is Usable. Interacting with Computers, 13 (2), 127-145.

Don Norman

Mogly wrote:

Doesnt Beauty and advertisment stick and interact together?


Leave a Reply


OK/Cancel is a comic strip collaboration co-written and co-illustrated by Kevin Cheng and Tom Chi. Our subject matter focuses on interfaces, good and bad and the people behind the industry of building interfaces - usability specialists, interaction designers, human-computer interaction (HCI) experts, industrial designers, etc. (Who Links Here) ?