Price Floors for Q4: A Practical Playbook for High-CPM Publishers

Why floors matter more in Q4

In a first-price auction, the highest bid wins and pays what it bid. That shifts the job of a floor. It is no longer mainly about lifting a second price up to a reserve; it is a signal of what your inventory is worth and a guard against bids that undervalue it. In the fourth quarter, when demand is at its strongest, a floor strategy built for spring leaves money behind. A floor set carelessly high, on the other hand, turns away bids that would have cleared, and unsold impressions do not come back.

Publishers in finance, insurance and personal finance feel this most. Their audiences attract high-value advertisers year round, and October through early December adds seasonal demand. Medicare's annual open enrollment runs from October 15 to December 7, which draws health insurance advertisers, and year-end planning content attracts financial services brands. When CPMs are already high, small floor mistakes are expensive in absolute terms.

Know the two floor systems you are running

Most publishers who use Prebid with Google Ad Manager are running two floor systems at once, and they interact.

Google Ad Manager pricing rules

Ad Manager's unified pricing rules set floors that apply across the open auction, private auctions, First Look, Open Bidding, header bidding and SDK bidding, and AdSense backfill. They do not apply to Programmatic Direct. Two details matter in practice:

• If two rules target overlapping inventory, the one with the higher price applies. Overlapping rules are a common source of floors that are higher than anyone intended.
• A network can have up to 200 pricing rules. That is enough for sensible segmentation, but it rewards a planned structure rather than rules added one at a time.

Prebid's Price Floors module

In the header, the Price Floors module lets you define floors by dimensions such as media type, size, domain, GPT slot and ad unit code, and pass them to bidders. It supports a floorMin baseline, enforcement settings that decide whether bids below the floor are dropped in the browser and whether deal bids must meet floors, and dynamic floors fetched from an endpoint. It also supports testing directly: skipRate randomly skips floors on a share of auctions, and modelGroups with weights lets you run several floor models side by side.

The practical point: your header floors and your ad server rules should tell a consistent story. If Prebid floors are much lower than Ad Manager rules for the same inventory, header bids can win in Prebid and then fail to compete in the ad server.

A Q4 floor playbook

1. Start from data, not a number. Pull the last 90 days of bid landscape data: bid counts, winning bid distribution and fill by segment. The right floor for a segment is somewhere below where most winning bids clear, not an average of all bids.
2. Segment where value differs. Geography, device, ad size and page type usually explain most of the variation in value. For finance sites, content type matters too: a retirement or insurance article often attracts very different demand than a market news brief.
3. Raise gradually. Move floors in small steps, weekly rather than daily, and watch fill rate and revenue per thousand pageviews, not just CPM. CPM almost always rises when floors rise. The question is whether total revenue does.
4. Keep a control group. Use skipRate or model groups so a portion of traffic runs without the change. Without a control, seasonal demand growth will make every floor increase look like a success.
5. Watch deal interaction. Decide explicitly whether deals must meet open-market floors. Many publishers exempt deals, then forget that some deals have their own low floors set in the SSP.
6. Check currency and net versus gross. If you sell in multiple currencies or bidders report net prices, verify that floors and bids are compared on the same basis.

Mistakes that cost the most in Q4

• Floors that never come down. Floors raised for November and December should be reviewed as soon as demand drops in January. A Q4 floor left in place in Q1 can cut fill sharply.
• One floor for all sizes. A floor that works for a large in-content unit may be far too high for a small mobile banner.
• Ignoring bid density. Floors work best where several bidders compete. On thin segments with one or two bidders, an aggressive floor mostly reduces fill.
• Changing floors and bidders at the same time. If you change both in the same week, you will not know which change moved revenue.

What to measure

Review results on the same day each week and compare against the same weekday in the control group, since traffic mix and demand swing between weekdays and weekends. Keep a simple log of every floor change with its date and scope, so anyone on the team can trace a revenue shift back to a decision.

Track revenue per thousand sessions as the primary metric, with fill rate, average CPM, bid rate and win rate by bidder as diagnostics. For finance and insurance publishers with long, research-heavy sessions, also watch pages per session: if higher floors lead to more empty slots or collapsed layouts, users notice.

The takeaway

Floors are the one pricing lever publishers fully control during the most valuable weeks of the year. Treat them like any other experiment: segment by real differences in value, move in small steps, keep a control group, and review them when the season ends. Publishers who work with a managed partner such as HBDR usually run this process as a standing weekly review through the quarter, so floor changes are measured rather than guessed.