A price floor is the lowest price you'll accept for an impression. Set it well and buyers pay closer to what your inventory is worth. Set it badly and you either give impressions away or leave them unsold. This guide explains how floors work in Google Ad Manager and how to approach them.
What a floor does
In an auction, the floor is the reserve price. Bids below it are rejected. If the highest bid is above the floor, it wins.
Floors help in two ways. They stop impressions selling for almost nothing when competition is thin, and they can encourage buyers who value your audience to bid higher. They hurt when they're set above what buyers will realistically pay, because then nothing wins and the impression goes unfilled or falls to lower-value demand.
Unified pricing rules
Google Ad Manager uses unified pricing rules to set floors. "Unified" means the same floor applies across Ad Exchange, Open Bidding and other programmatic demand that competes in Ad Manager's auction. Before these rules existed, publishers could set different floors for different buyers, which created ways to game the auction. Google removed that option in favour of one floor per rule.
A pricing rule combines two things:
- Targeting. Which inventory the rule applies to, for example certain ad units, sizes, countries or devices.
- A floor price. The minimum price for that inventory.
When several rules match the same impression, Ad Manager applies them according to its own precedence logic. Keep your rules simple enough that you can predict which one applies.
How to set floors sensibly
There's no universal right number. A floor that's perfect for a UK desktop leaderboard will be far too high for a mobile banner in a low-spend market. A practical approach:
- Start from data. Look at what each segment actually earns: country groups, device types, ad sizes.
- Segment where prices really differ. Country and device usually matter most.
- Set conservative floors first. Close to the lower end of what buyers already pay, not the top.
- Change one thing at a time. Adjust a floor, wait long enough to see the effect, then decide.
- Watch revenue, not just CPM. A higher eCPM means nothing if fill drops so far that total revenue falls.
The trade-off: price vs fill
Every floor is a trade between price per impression and how many impressions sell. Raise the floor and the average price of what sells goes up, but fill rate goes down. The right floor is where total revenue is highest, not where the CPM looks best.
| Floor | What usually happens |
|---|---|
| Too low | High fill, but some impressions sell far below their value. |
| About right | Strong fill, buyers pushed to bid nearer their real value. |
| Too high | Good-looking CPMs on what sells, but many impressions go unsold. |
Common mistakes
- One floor for everything. Your traffic isn't uniform, so a single floor is wrong almost everywhere.
- Copying someone else's numbers. Floors depend on your audience, not theirs.
- Setting and forgetting. Demand changes through the year. Floors that worked in spring can be too low in Q4 and too high in January.
- Too many overlapping rules. If you can't tell which rule applies, you can't learn from changes.
- Judging too quickly. A day of data is noise. Give changes time.
What about automated floors?
Ad Manager also offers optimised pricing features that adjust floors automatically based on bidding behaviour. They can work well, especially on large volumes of traffic. It's still worth keeping sensible manual rules as a baseline and watching total revenue when automation is switched on.
A simple way to keep track is a short change log: the date, which rule you changed, the old and new floor, and what happened to revenue, fill and eCPM over the following two to three weeks. It sounds tedious, but after a few months it tells you more about your inventory than any general advice can. It also stops the same experiment being repeated by someone who didn't know it had already been tried.
When you do review results, compare like with like. A floor raised in late November will look brilliant because Q4 demand is rising anyway. A floor raised in early January will look bad for the opposite reason. Compare against the same period for segments you didn't change, so you can separate your change from the season.
And don't forget invalid traffic. Buyers already discount inventory with poor quality. Raising floors on that traffic rarely works, because buyers simply stop bidding. Cleaning up traffic usually does more for prices than any floor setting.
Rather not tune floors yourself? Setting and reviewing pricing rules is part of the ad operations Bidora handles for the sites we manage. Talk to us about joining Bidora MCM.