The question of whether one should run A/B tests (a.k.a online controlled experiments) using one-tailed versus two-tailed tests of significance was something I didn’t even consider important, as I thought the answer (one-tailed) was so self-evident that no discussion was necessary.
