A/B testing for consumer-facing user experience is best when you have a hypothesis of what will make the experience better and then testing it, rather than trying random shit and throwing paint at the wall.
For instance, the typography is pretty poor on the site. One of the first hypotheses I would validate are things around improving legibility.