One of the stranger things we keep seeing in Meta is the same creative behaving completely differently on different runs.
Same image. Same message. Similar audience. One outing does very little. Another suddenly works.
That makes creative testing slightly less tidy than we’d all like it to be.
The early signals matter. Who sees it first matters. What Meta learns from those first interactions appears to matter. And sometimes a perfectly good piece of creative just gets a bad first outing.
So we’ve become less willing to permanently bin something after one weak result.
Sometimes we change it.
Sometimes we just run it again.