Questions I have about Multivariate Testing:
Would it be of value and/or possible to test variables in combination - IE, headline A with copy C makes more sales than headline B - even though headline B makes more sales with all other possible combinations?
My other question has been addressed - would it be possible and of value to test/track where the visitor is coming from? Answer is here
MuVar is getting updated - both improved and having bugs fixed.
James Brausch has used one of my suggestions, but has a really good reason for not using the other one - namely that the HTML code can get messed up. He asks if I used Firefox - I did. I also used Microsoft’s Internet Explorer. I also tested with multiple computers. So I would bet that the problem is a php configuration on my server. I haven’t tracked it down yet. I will post here if/when I get it figured out.
For anyone else trying to figure it out, it seems that on my server (and also some - very few - others,) once a cookie is set, it doesn’t get overwritten or reset. This is not good when entering new information that gets passed in a cookie. But my fix - using hidden fields in a form - is not good either because in the posted data HTML will get messed up. One possible solution that I haven’t tested yet is to mangle the data in the hidden fields so that there is no html there, and then after they are posted, unmangle them. PHP includes to functions that do this easily, base64_encode and base64_decode.
I probably won’t be able to test this idea or look for other solutions before next week. (I try to observe the sabbath on Sunday, and I have procrastinated taxes until the last minute) but if I find anything I will let everyone know.