...are rarely websites that convert as well as unpretty ones.
If the goal of your site is to position you, tell a story, establish your good taste and make it clear what sort of organization you are, then pretty might be the way to go. And you can measure the effectiveness of the site by how it impresses those you seek to impress, by its long-term impact.
But it's a mistake to also expect your pretty website to generate cash, to have the maximum percentage of clicks, to have the most efficient possible funnel of attention to action.
There's always been a conflict between the long-term benefits of beauty in commerce (in architecture, in advertising, in transactions) and the short-term brutality of measurement and direct response.
It's worth noting that conflict in advance, as opposed to vainly wishing you could have both optimized. You can't. The smart marketer will measure how much direct response it's costing to be beautiful, or how much storytelling is being sacrificed to be clicked on. Not both.
[A few readers asked me to expand on this idea: It turns out that in most encounters, the worldview of people who are likely to sign up, 'like', share, click, act and generally take action instantly is not the same worldview of people that convert into long-term, loyal customers over time. Take a look at the coupons in the Sunday paper, or the direct mail pieces that show up in your mailbox, or the websites that are optimized for click/here/now.
Unattractive high-response sites aren't usually the result of a lack of taste or talent on the part of the designer, they're optimized for one worldview.
The design that you and I might see as non-beautiful is in fact a signal to one group of people just as much as it is a turn off to the other group. My argument is that you can optimize for one group or the other, but you can't likely optimize for both.]