Well, the ease of implementing A/B testing should keep patio11 happy... I've noticed he seems almost religiously fanatical about it, and probably for good reason! Looking back at a quick search, it seems he's been doing it for several years...
Hopefully it will go further than what he's anticipating: And yet when we check back on who A/B tests here in February 2013, do you know what we're going to see? A few articles about it, with wise comments like "A/B testing can only get you to a local maxima" and "Didn't we read an article just like this on Smashing Magazine in 2006?", and fairly few companies who actually have someone do it every week, even among the savvier folks.
It is also good for staging new version of apps that you only want some users to have access to. I have been doing this manually by having the new version called 'staging' and setting it in the user session to redirect.
You could also use this for browser testing, where your test client sends a header that will redirect it to a clean datastore with the new version
Hopefully it will go further than what he's anticipating: And yet when we check back on who A/B tests here in February 2013, do you know what we're going to see? A few articles about it, with wise comments like "A/B testing can only get you to a local maxima" and "Didn't we read an article just like this on Smashing Magazine in 2006?", and fairly few companies who actually have someone do it every week, even among the savvier folks.
http://news.ycombinator.com/item?id=3639365
http://news.ycombinator.com/item?id=3646669
http://news.ycombinator.com/item?id=1404083
http://news.ycombinator.com/item?id=764508
http://news.ycombinator.com/item?id=484668