Google’s webmaster tool video dealt with this question in a recent video.
When Google encounters a page without links it judges it by the content and the keywords it finds. The first instance of the keyword indicates the subject matter then several repetitions confirms that page is about that subject matter. However, there comes a point where the repetition is judged as keyword stuffing and a negative effect occurs. If the page is about a very niche topic or phrase then it has the potential to rank better for that non-competitive keyword than a highly searched phrase.
A/B testing is an easy way of comparing two versions of a page to know which converts better. Multivariate testing on the other hand compares multiple changes of possible web design and can be expensive and complex.
How does do the two compare
A/B requires little development and can be implemented by SMB’s with limited budget and technical knowhow. You can get a crowd source web page design at very low cost and use Google analytics to setup the tracking parameters.
A/B compared to multivariate testing – http://www.smashingmagazine.com/2010/11/24/multivariate-testing-in-action-five-simple-steps-to-increase-conversion-rates/