Ask HN: Do you worry about your a/b tests influencing targeting?

Do you consider that setting up a profound test (e.g. price model change) may influence the traffic that Facebook sends to you? It can be reasonably thought that conversion rates you see may be affected by the split (because half the users see an alternative option) so even if you decide against the alternative the targeting is already borked. I've heard this theory from some people and I wonder if it can be true.