What the author describes is the equivalent of premature optimisation for design. I regularly advocate a "get something up then refine" approach to our clients

What the author describes is the equivalent of premature optimisation for design. I regularly advocate a "get something up then refine" approach to our clients but it's difficult. They don't want to bear the extra costs and usually don't understand the need. And when you present that first comp to them and it doesn't look perfect, they will balk, first round or not. Startups are more interested in the approach, but then they really don't like the extra costs :-) Secondly, as other people have pointed out, design isn't just aesthetics. It is also, and primarily in the case in web design, about functionality and usability. If you have hired a good design agency are only using them to "make things pretty" then you are underutilising them. Finally 20% can mean a lot of value. In some cases it's the "winning edge" over your competitor (aka life of death) and in others it's 20% of revenue. Admittedly spending huge amounts on a site that only nets a few thousands a month is silly but if the site is big enough it makes perfect sense.