[Wikimedia-l] [Wikimedia Announcements] Wikimedia Foundation Report, May 2012

[James Salsman]

I also want to say something good.  I think the fact that the
fundraising team is using multivariate analysis instead of simple A/B
testing now is beyond good, it's just spectacular.  A/B testing was
excruciatingly slow, and this is a huge advance.  I hope it means that
all the banner text suggestions from 2010 will be tested.  There was a
very large variance in the performance of those banner text
submissions, suggesting that the best have yet to be found.

Ref.: http://meta.wikimedia.org/wiki/Fundraising_2012/We_Need_A_Breakthrough#16_May.2C_2012:_Multivariate_landing_page_testing.21