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 tim

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.