[boucher]

You are skeptical because you believe that this is a data problem. He believes (as do I) that it is not a data problem.

Making more people click your button is not, on its own, good design. There's more to aesthetics than conversion rates. You can argue aesthetics aren't important, but it's simply not the same subject as the kind of data collecting A/B testing that Google does.

[tptacek]

I have a hard time understanding how people can reject the notion that design is a data problem. I know I might be wrong about this. Random thoughts:

If your design doesn't accomplish a business objective, it may be better, but it is irrelevant to the business.

If data indicates your aesthetic doesn't increase a metric, then ultimately the only rational decision to make is to minimize the cost of your aesthetic. All things being equal, you take the equivalently successful design that imposes the least costs.

If the data indicates that your design decreases a metric, then ultimately the only rational decision is to kill the design.

There are also probably non-obvious cost factors. A much better Google SERP design may be more aesthetically strong, and may not hurt metrics --- it may even marginally improve metrics --- however, if the end result of that aesthetic is that every other page at Google needs to cost more to cohere with the SERP, then, again, it may be irrational to improve the design.

Companies spend a lot of money on design. Companies tell themselves a lot of things about design. Somebody's whole career is predicated on the idea that the lobby of a class A office building in the Chicago Loop needs to be just-so. In the building I'm in now, somebody spent several million dollars to improve the aesthetics. I think that probably has more to do with vanity than it does with business.

Corporate annual reports are a marquee design project. Companies spend serious dollars on them. Berkshire Hathaway does theirs in Word, set in Times New Roman. They seem to be doing OK.

Sell me on aesthetics more? I don't see it.

[fjabre]

Design can be a powerful weapon if done right - just look at Apple..

However, it's very difficult to achieve Apple's near perfect blend of aesthetics & function so I think most people give up and build anyway, making something perfectly functional i.e. Facebook or Gmail, but not awe-inspiring.

[tptacek]

Apple's design is also more data-driven than it appears --- although it's clearly not as coldly rational as Google's.

[boucher]

Data driven from an individual collection and observation perspective, perhaps. Apple doesn't AB test its products.

[eagleal]

I think everyone misses the point of the design choice of Google, and many others. By tests I can confirm that a mathematical (analytical) approach to design, it's better (read processed better) than the majority of purely visual ones.

I can repeat this all the day: Google search, it's one of the most well designed web applications in the entire web.

GMail? Not so much. I have done some (NDA research) email UI/UX "design" at my company, and by tests (just the second revision), I reduced the lookup (finding items), for first-time computer users by 57% over gmail, by 41% over hotmail, and by more than 700% over ymail (2008). This because of a more cautionary choice of color of the controls. This was the goal.