2 ms·

Would you rather they change things for everyone at once without testing?

• aeinbu 8mo ago
  That is not the only other alternative. You can do A/B testing splitting up your audience in groups, having some audience use A, and others use B - all the time. I think the article’s author is frustrated over sometimes getting A and at other times B, and not knowing when he is on either.