Forum Moderators: not2easy
Has anybody tried the technique - which has a number of uses - with website sales copy?
I know my site can sell better, but i'm always nervous to try new things incase it makes matters worse! Sure I can track over time, but I don't want to lose sales! A/B Split could work quite well - I might still lose sales, but hopefully not as many, whilst possibly seeing a big improvement.
You can use a cookie to try and make sure that returning visitors see the same version each time.
Any thoughts?
Run two ads, wait two days, drop the worse ad. Create new ad as variation on the remaining ad. Run two days... and so on.
Should result in a steady improvment, and I think the "loss" in sales of havign a possibly worse performing second ad are negligible if the second ad is a variation on the good performing ad.
it could even be a A1/B1 A2/B2 split, where you have to distinct strategies competing, and each has two variations on the strategy to evolve the strategy. Regularly you drop the worse variant, and less frequently you drop the worse strategy and create a completely new, un-related one.
SN
We have two versions of our site served up on two identical servers. When no testing is taking place, the two sites are exactly the same.
Then, when I want to test something, I tweak a page one one server only, and let the two pages run head-to-head for a week or so to be able to detect statistically significant differences between the two pages.
The trick is, just assign users on a round-robin fashion to each server so you get 50% of your traffic seeing one page and 50% seeing another. If you're contemplating big changes, just dial down the amount of traffic going to the "changed" page and let it run for longer ... that way you'll do less damage (lose fewer customers) if the change isn't a good one.
You can do this for copy/headline tweaks all the way to total design changes.
