I've never done an A/B test on typeface, specifically, but I've made a few over the years on e.g. font sizes. No major results to report. My intuition is that I'm extraordinarily skeptical that this would matter and largely think the only people capable of perceiving strong differences in the character of a typeface are people who do that professionally plus Thomas. I'm prepared to be wrong about that, though, as six years ago I would have told you that colors could not possibly meaningfully impact conversions and that's just catastrophically wrong.