A publisher floor test found revenue peaked at $1.20, not the $0.80 in use

A publisher floor test found revenue peaked at $1.20, not the $0.80 in use

A publisher set floors by intuition. A structured elasticity test on first-price auctions found the revenue-maximizing floor sat well above their default.

Floor price (the minimum a bid must clear to win; in first-price it also pressures bidders upward toward it).

1. They ran a floor ladder on one placement: rotating slices at $0.80, $1.00, $1.20, $1.40, $1.60.
2. They measured fill rate (share of auctions that resulted in a paid impression) and revenue per thousand auctions at each level.
— $0.80: 95% fill, $0.91 RPM.
— $1.20: 88% fill, $1.14 RPM.
— $1.60: 64% fill, $0.98 RPM.
3. Revenue per thousand peaked at $1.20, where modestly lower fill was outweighed by higher cleared prices.

Evidence: moving the placement's floor from $0.80 to $1.20 raised revenue 25% per thousand auctions. In first-price, the higher floor also nudged bidders to bid up rather than shade down to a low floor.

Why it matters: in first-price auctions the floor is an active pricing tool, not just a safety net. The revenue-maximizing floor is found by testing the elasticity curve, and it is often higher than the fill-maximizing one.