How many offers should I test at once?

How many offers should I test at once?

Q: Should I throw 10 offers into a rotation and let the best win, or test one at a time?

A: Neither extreme works. Ten offers at once splits your traffic so thin that none reaches a meaningful sample — you'll "learn" from noise. One at a time is too slow and you can't compare fairly.

A practical approach:
— Test 2-4 offers head-to-head on the same traffic at the same time, so geo/time/source are equal
— Give each enough volume to hit a real sample (see the 1-3x payout rule)
— Kill the clear losers fast, then run the top 1-2 deeper
— Keep the variables identical — only the offer changes, or you won't know what caused the difference

Short version: split-test 2-4 offers at once on identical traffic, not 1 and not 10. More Qs? Drop them.