calling it “ad fatigue” and moving on isn’t a diagnosis — it’s a shrug.
what’s actually going on: Meta’s Andromeda algorithm hands a winning creative more impressions, those extra impressions burn through your audience’s patience, CTR drops, CPL climbs — usually inside 2-3 weeks on a mid-budget account. nothing broke on the algorithm’s end. your ad simply ran out of fresh eyeballs to persuade.
so “make more ads” isn’t the fix. having a replacement ready before the current winner decays is.
the 3-layer structure I run on every account:
- concept testing — 3-5 genuinely distinct concepts (not palette swaps), one ad apiece, ABO campaign, equal budget per ad set. this isolates the idea itself, not the execution around it.
- variation testing — whichever concept wins gets 3-4 execution variants: hook, format, opening frame.
- scaling + rotation — the winner moves to a CBO scaling campaign while 2-3 backups keep warming on a small testing budget, so nothing goes dark once fatigue sets in.
kill criteria carry as much weight as the test structure itself: pause anything that hasn’t proven itself within 48-72 hours or $50-100 in spend, whichever hits first. the CPL curve will tell you before a chart confirms it.
stick with ABO through the testing phase — CBO tilts toward early performers and skews the read before you’ve learned anything real. bring in CBO once you’re scaling a confirmed winner, not before.
for B2B accounts: test the value proposition before the visual polish. problem-aware framing versus benefit-led framing shifts results more than a sharper thumbnail ever will.