Would you say that David Lynch is a suboptimal director? His movies aren't enormous hits. He hasn't won for Best Movie. But those aren't his goals. Similarly, sometimes the people developing a site have more at stake than maximizing profits. Sometimes they want to make gorgeous web pages or gorgeous processes. And there's no optimal procedure for that.
Some people look at it another way. The 37s team obviously treats design as a sort of art rather than an exercise in optimization. As it happens, they make better products (in my opinion) than the people who go the other route. So they serve as a sort of inspiration.
That's my point. The argument is not entirely meaningful, because there are two things happening at once, and you can measure the one but not the other.
See, saying that sales are drying up is a completely different thing. You'd be stupid not to change from that. I thought your question was about the intuition-based design process, period, because you can design intuitively and NOT lose all your profit. The 37signals design isn't making sales dry up, for instance, even if they didn't run tests on it. It's subjectively a much better page.
Obviously there a "don't do stupid stuff" rule in place, where if intuition is very, very wrong you slap the intuitor and start again.