"Creatives fatigue, so kill the winner and find a new one" — usually wrong
When ROAS dips, the reflex is "creative's tired, scrap it." Half the time you're scrapping a healthy ad for a problem somewhere else.
The setup
Gambling offer, my best creative doing 2.6x for three weeks, then drifting to 1.9x.
The move
Before killing it, I segmented the decline by placement and frequency instead of blaming the asset.
The numbers
Frequency on my core audience had hit 6.2. The creative still did 2.5x on fresh users — the drop was entirely from over-served existing ones. Audience exhaustion, not creative fatigue. Expanding the audience by 3x restored 2.4x ROAS with the same "tired" creative. The asset had another $9,000 in profit left in it.
The lesson
Declining ROAS on a proven creative is usually an audience-size or frequency problem, not the asset. Killing the creative throws away your best tested variable.
What I'd do differently
Before retiring any winner I check frequency and audience saturation first. New creative is the expensive answer to a question audience expansion often solves for free. Numbers illustrative; the misdiagnosis is everywhere.
Arb Files
@ArbFiles