We are busy getting a revised page in place to replace the current page which displays categories on our widget site. The page will have a different design and layout, and display different information to the user. The objective is to run the current category pages and new category pages simultaneously over a period of 2 weeks and monitor conversions. We will then use whichever page performs better, provided the results are of statistical significance.
My question is this - we will have the same URL showing 2 different pages of content over the 2 week testing phase. Will this have any negative affects on our rankings? Could Google perhaps flag us for content that changes too often or anything along those lines?
IMO, the best option is to split test away from organic search of any kind. A reliable option is to use pay per click traffic, and robots exclude any pages you use for testing. There may be differences in audiences for paid search, but it's about a good a benchmark of performance as I think is available.
If you're split testing without restricting the variables (e.g. by testing with all your site's traffic) then potentially your test won't give statistically reliable results anyway.
I must say, that is quite a cunning method of testing the pages. The only issue would be having to set up so many paid ads to get enough data for the sample set, could be time consuming.
I have just been reading up more on Google's website optimiser and see that it offers both A/B and multivariate testing. I think this will be ideal for what we need unless I'm missing something? I'm guessing it may well provide more data than just conversion tracking and we would not have to worry about SEO implications as consistent data would be served to search engines.