Myth: a native channel showing strong ROAS is obviously incremental.

Myth: a native channel showing strong ROAS is obviously incremental.
Wrong — and one brand learned it for $0. They had a native retargeting feed reporting a 6.2x return on ad spend, the best line in the deck. Then a public-holdout test: randomly suppress the ads for 50% of the retargeting pool.

The suppressed half converted at 5.8% over 30 days. The exposed half: 6.1%. The "6.2x ROAS" channel produced a real lift of 0.3 points — almost entirely re-billing purchases that were already coming.

They reallocated the budget to prospecting, where a holdout actually showed lift, and total revenue rose 8% on flat spend.

ROAS counts conversions that pass through the pixel, not conversions the ad caused. Without a holdout, your best-performing channel might just be your best-decorated one.