On Thursday, I spent a while talking to a friend who does nothing but multivariate testing and website optimization. He told me that it is not uncommon to capture fully 3/5 of the final conversion improvement for a client in the first cut - before any tests are run (which then get you the rest of the way).
There are two factors driving this: Firstly, most companies full conversion funnels are just plain terrible - even many established SEM driven lead-gen businesses. The money used to be so easy that efficiency was just not critical. The problems are bubbling up now because competing PPC bids are getting higher and competing strategies are getting more sophisticated. Secondly, newsflash, a lot of best practices in landing pages and conversion are, in fact portable. I found this 3/5, 2/5 back-of-the-envelope split quite interesting, although it clearly has limited applicability to other less-experienced practitioners.
I am in Copenhagen all this week, so I owe you, my public :) some photos, et cetera - expect it!