The idea behind an A/B Test is to test designs, roll them out and do it all over again. So iterate and never stop. Because you will never find the winner on an A/B Test.

Iterate the design and test again. The design process never ends!

50-50 doesn't just mean that they are both good.

For example if I want to A/B test a design with signup pre and post an action, my hypothesis might be that a pre signup screen will reduce conversions. If my results are 50-50 that shows that there is no significant drop off by forcing a user to share their details. And therefor that should be the pushed out design, as it collects more data and doesn't have any more dropoff whilst doing so.

So it depends on what your hypothesis behind the A/B test was.