Nobody outside Google can change a Google Ad Manager price floor during a live auction. The vendors who say otherwise are either pushing delayed updates through the GAM API or moving floors inside your Prebid wrapper. What actually raises revenue is duller and more reliable: put what buyers bid next to what your floors ask, find the gaps, and fix the rules yourself on a schedule.
Below: how price floors work in GAM in 2026, where the money leaks, how to find the leaks with the reports you already have, and how to run a floor review that a two-person ad ops team can sustain.
What is a price floor in Google Ad Manager?
A price floor is the minimum CPM a buyer must clear for an impression to be sold. In Google Ad Manager the floors for programmatic demand are set through Unified Pricing Rules (UPRs): one rule can target ad units, sizes, devices, geographies, key-values, or a combination, and a fixed floor applies to every buyer that rule reaches. Since Google moved its exchange to a first-price auction in 2019, the floor is also the only signal buyers get about what you think the impression is worth, so it shapes how they bid, not just whether they win.
Two things trip people up:
- Two floor systems, not one. UPRs govern the ad server. Your header bidding wrapper has its own floors, set through Prebid's Price Floors module or per-bidder parameters. If the two disagree, buyers see one floor in the wrapper and another when the bid reaches GAM, and the impression can go unsold while both systems report "working as intended".
- Floors cost fill. Every cent you add to a floor removes some bids from the auction. A floor is a bet that the remaining bids are worth more than the ones you lost. The bet has to be checked with data, per segment, on a cadence.
Why manual floor management breaks down
The typical setup is a few dozen UPRs written at different times by different people. Over a year, the stack around them changes: new bidders, new sections, a redesign that moves the mid-article unit, a currency swing. The rules do not change with it, and two leaks open:
- Floor gaps: the rule asks for less than buyers are already willing to pay. Bids clear just above the floor, and the price discovery you were supposed to get from a first-price auction never happens.
- Fill drops: the rule asks for more than the market will pay on that segment (weekend mobile traffic from a low-CPM country, for example). Impressions go unfilled or fall to house ads while the rule sits there looking prudent.
Neither leak shows up as an alarm. Revenue drifts a few percent, and the cause is spread across twenty rules.
Can a third-party tool change GAM floors in real time?
No. Google's ad server does not let external code intervene in a live auction. A vendor can:
- update UPRs through the Ad Manager API on a schedule (minutes to hours), which is batch optimization, not real-time bidding logic;
- change floors in the Prebid wrapper before the ad request leaves the page, which affects header bidding bids but not what AdX and Open Bidding see in GAM;
- or rely on Google's own Target CPM option inside UPRs, where Google, not the vendor, moves the effective floor to hit an average CPM you set.
All three can be useful. None of them is "our algorithm changes your floors millisecond by millisecond", and a vendor who describes it that way has told you how carefully to read the rest of the pitch.
How to find floor gaps with the data you already have
You need two datasets side by side, at the same grain: what buyers bid and what the rule asked for. Work per ad unit, device, and country, because that is the grain at which the gaps hide.
1. Pull the bid side
- Header bidding: your Prebid analytics adapter, or page-level bid logging, gives every bid per bidder, ad unit, device, and geo, including the ones that lost. The distribution of bids matters more than the average: a unit where 40% of bids sit within a few cents above the floor is telling you the floor is anchoring buyers.
- Ad Exchange and Open Bidding: GAM's bid data reports (where your network has them) include a bid rejection reason dimension. "Below floor price" volume by rule is the fastest way to see a floor that is choking demand.
2. Pull the rule side
Export your UPR list with targeting and floor values, and map each rule to the segments it actually reaches. Rules overlap, and GAM applies the higher floor when they do, so the floor a segment really has is often not the one written on the rule you remember.
3. Compare, segment by segment
Two patterns are worth acting on immediately:
- Gap up: the median winning header bid on a segment sits well above the UPR floor (a common find is 25% to 40% on desktop news sections). The floor is not doing anything; raise it in steps and watch fill.
- Gap down: a segment shows high unfilled or house-ad share and a pile of "below floor" rejections. The floor is too high for that traffic; lower it or split the rule so that the premium segment keeps its floor and the rest can clear.
A worked example: a Finance section where Prebid eCPMs consistently clear 35% above the UPR floor on mobile is leaving money on the table on every impression that sells; the same section's weekend traffic from a low-CPM geo may be sitting unfilled under the same rule. One rule, two opposite fixes.
How to run a floor review your team can sustain
- Weekly, 30 minutes: look at the ten segments with the largest revenue and the ten with the largest unfilled share. Anything that moved more than 15% week over week gets a note.
- Change one thing at a time. Raise or lower a floor by 10% to 20%, not 50%, and hold it for at least a week of comparable traffic before judging. Buyers' algorithms take days to re-learn a floor.
- Keep a change log. Rule name, old floor, new floor, date, the reason, and the number you expect to move. Half of floor management is being able to answer "who changed this and why" three months later.
- Reconcile the wrapper quarterly. Compare Prebid floors with the UPRs that cover the same inventory and remove the disagreements.
- Use Target CPM deliberately. It is good at holding an average on stable inventory and bad at protecting a premium segment inside a broad rule. Do not switch it on network-wide and walk away.
Where Optimon fits
Optimon connects to your Google Ad Manager and Prebid data and does the joining above for you, every day. It flags the segments where bids and floors have drifted apart, tells you what moved and why, and each floor suggestion links to the exact pricing rule in GAM so your team makes the change with the numbers in front of it. It does not change floors behind your back, and it does not pretend to bid in the auction. Globes manages its AdX floors this way; the case study has the numbers and the operational savings.
FAQ
Can an external platform change my GAM unified pricing rules in real time?
No. Google Ad Manager does not allow third-party code to alter a live auction. External tools can update rules through the API on a schedule or adjust floors in the Prebid wrapper. Google's own Target CPM setting is the only mechanism that moves floors dynamically inside GAM.
Should my Prebid floors match my GAM unified pricing rules?
They should agree on the same inventory. A wrapper floor below the UPR lets a header bid win in the wrapper and then lose in GAM; a wrapper floor above it throws away bids that GAM would have accepted. Reconcile them at least quarterly and whenever you restructure rules.
How often should price floors be reviewed?
Weekly for the largest segments, with individual changes held for at least a week before you judge them. A full rule-by-rule audit two to four times a year catches overlaps and rules nobody remembers writing.
What is a floor gap?
A floor gap is the distance between what buyers are willing to pay for a segment and what your floor asks. When bids consistently clear well above the floor, the floor is too low; when a segment shows heavy "below floor" rejections and unfilled impressions, it is too high.
