08/20/2026
Meta says 12 ROAS.
First-party data says just under 6.
Still good.
Also a completely different creative decision.
That’s why creative-level reporting gets dangerous when you stop at the platform number.
In one view, Meta says a specific creative is driving a 12 ROAS. Store-side data says it’s closer to 6.
You probably don’t kill that ad.
But you also don’t scale it like a 12 ROAS winner without checking the rest of the story.
Then it gets more interesting.
Switch the view to new customers.
Meta says the same creative is getting about a 2 ROAS.
First-party data says 18.6.
That’s not a rounding error.
That’s the difference between:
“This ad isn’t working.”
And:
“This ad may be speaking directly to new customers.”
If you’re looking at blended performance, or the wrong customer segment, you can kill creative that’s doing exactly what you need it to do.
Especially for acquisition.
This is why creative reporting needs more than 1 ROAS column.
Before you pause, scale, refresh, or brief the next batch of ads, look at:
• Meta-reported ROAS
• First-party or store-side ROAS
• New vs returning customer performance
• Revenue driven by the creative
• Order volume and order spikes
• Trend over time
• Product featured vs creative format
Those order spikes matter too.
If a creative launched, spiked, then slowly declined, ask why.
Was it tied to a seasonal moment?
A Labor Day sale?
A product trend?
Or is the creative decaying?
Because the fix changes depending on the answer.
If the product is fading, you test a different product.
If the format is wearing out, you test a new angle or hook.
If it was seasonal, you don’t overreact to the decline.
Analyze first.
Optimize second.
Before you change creative or move budget, do you know which customer type that ad is actually working for?