Focus your product, launch it aggressively, sell it aggressively, no half-assing. Anything that gives you an excuse to half ass is bad. A/B Testing gives you that excuse.
Focus on one product, iterate on it, and if it doesn't work, try another focused product.
When you A/B Test without volume or some decent amount of traction, whatever results you get will be skewed and won't tell you things accurately anyway. And if you're now worrying about managing two products or even more, your focus is scattered and will dilute the quality of each of your products that you maintain. And that minor dilution could be that extra inch you need to achieve product market fit.
If you're A/B testing chances are you aren't ruthlessly prioritizing what to do enough. You're trying to hedge risk by making a worse product. Instead make a bet, eat the risk you're about to take and put it on the table. Better to do it and fail.
When A/B Testing, you are also creating a mental branch. And as you create more and more branches, you forge different paths for your product which makes everything confusing. It becomes hard to keep track of and sync changes (despite any attempt at maintaining great engineering practice). (This is why its so much faster for people to push to master when using Git and not worry about merging branches)
So when you're starting out, don't A/B test, just ship and iterate linearly.