+3 votes
Something we occasionally notice while running creative tests is that legacy/historical creatives get most of the spend(and strong performance) because these have social proof(likes/comments/views) that FB tends to favor -> and in these cases, new creatives dont always perform well when benchmarked against 'control' creatives when we run creative tests(either multiple ads in a single ad set, or 5-6 different concepts in different ad sets).

when you're running your first round of tests how do you account for the fact that the older creatives are doing well mainly because of social proof? in other words, how do you avoid killing a potentially high-performance creative that may be starting slow just because it has no social proof?
by (2.3k points)

1 Answer

+1 vote

To me, the only I have ever been able to solve this is through having a dedicated testing accounts that target a percentage of your core user group and creating new campaigns / adsets with new ad creatives.

Even if the overall performance is worse, all I care about at this stage is relative performance.

Using an ad study / holdout group on account level may make this even more feasible. https://developers.facebook.com/docs/marketing-api/reference/ad-study/

That way you can have 90% of the world in your regular campaign setup and 10% dedicated for testing.

by (370 points)