fail fast, fail often

I find this impressive. In less then a year, Google established its logo, and kept it for almost a decade without any change. [luckily, they figured out pretty early that they should lose the exclamation point. yahoo! unwittingly kept it - and now look where that got them] The point, however, is not in the logo: rather, it's the process of iterations that made (and still) makes google one of the strongest brands today. No wonder, too, that there it is in the company of other digital brands, like Amazon, Netflix, eBay, etc. While the question of why those brands are successful, and more importantly, how to imitate them eludes their offline counterparts (and is a million dollar puzzle for agencies who build brands' digital campaigns), the answer may be something really simple: test before you launch. [yea i know, sounds techy but in fact it's not]. The old advertising thing is "50% of my advertising fails ... " In my opinion, that's a terrible success vs. failure rate. [and makes one wonder why (and how?) this industry lived as long as it did, the way it did]. The thing is that anything (any solution, idea, etc) is very likely to have a 50% success rate when tested the first time. Google's (or Amazon's & Netflix's) solutions prob have the same likelihood of succeeding as any ad spot or a brand campaign, but then they get - through a number of iterations & feedback - repeatedly tested so when they are launched, this rate goes up to something manageable, like 90-95% (altho that's also prob still not high enough). So the question is: what makes ad agencies so sure - in absence of any iteration - that the products they deliver are going to succeed? (and what makes them to continue to be so sure altho they are proved right only in 50% of cases, a rate that clearly borders randomness). The thing that makes this kind of thinking even more appealing is that digital agencies - which, in fact, can iterate - mostly choose not to, but rather "test" their product by measuring impressions&click-through rates only after it's already launched. If success of digital brands resides in their process [summarized as: try as many things as you can, get a  diverse feedback, fail quickly, focus on what works], then why not imitate it? Good news: some smart places are already doing it.