We ran a coupon holdout.
By gabriel-caceros-528077bb · September 10, 2026 · Curated by George's Blog
We ran a coupon holdout.
The dashboard hated it.
The P&L did not.
A household brand wanted 15% off heading into September "to keep momentum."
CVR on coupon ASINs had always jumped.
That jump was being used as proof the coupon created buyers.
We held the coupon off 30% of comparable ASINs for 21 days.
Same creative.
Same bids.
Same category.
Coupon ASINs:
→ CVR up 18% versus their own baseline
→ ACOS down 4 points
→ Ad-attributed orders up
Holdout ASINs:
→ CVR barely moved
→ Orders came in 14% lower than the coupon group
→ Contribution per order $3.80 higher
When we lined them up, most of the coupon "lift" was buyers who were going to purchase anyway, plus a thin slice of true incrementality we measured at about 4%.
The coupon was not a growth engine.
It was a margin transfer with a conversion trophy.
We kept a smaller offer on the two slow variants that failed the holdout test the other way.
Hero SKUs went back to full price.
Monthly contribution up $8,900.
Sales volume down 6%.
The holdout was not pretty in the ads console.
Coupon ASINs "won" every efficiency screenshot.
Finance won the month.
If your coupon test has no holdout, you are measuring CVR with a gift attached.
Of course CVR rises.
The question is whether the order would have happened at $0 off.
If you cannot hold something out, you do not know if the coupon is creating a customer or paying your existing one to use a code.
Hold out a comparable set for 21 days.
Live with the uglier screenshot.
Then keep the code only where incrementality shows up.
Incrementality is not a vibe.
It is a group you were willing to leave un-discounted.
#amazonads #amazonppc #amazon #ecommerce