Price Floors in a First-Price World: A Guide for B2B and Niche Publishers

B2B and niche publishers have an unusual yield problem. Their audiences are small, specific and often extremely valuable: IT decision-makers, procurement managers, clinicians, engineers, small business owners. A handful of advertisers will pay a lot to reach them. Most of the open market will not bid at all, or will bid as if the audience were generic.

Price floors are the main tool for protecting that value, but they behave differently than they did when many auctions were second-price. This guide explains why, and how to set floors that raise revenue instead of just lowering fill.

How first-price changed the job of a floor

In a second-price auction, the winner paid roughly the second-highest bid. A floor acted like a phantom second bidder: if only one real bid came in, the floor set the price. Floors could lift revenue significantly on thin auctions, exactly the kind B2B publishers see.

Programmatic has largely moved to first-price auctions, where the winner pays what it bid. Google completed its move to a unified first-price auction in Ad Manager in 2019, announcing the full rollout that September and introducing unified pricing rules to set floors across indirect demand. Header bidding was already first-price for most exchanges.

First-price auctions brought bid shading. DSPs now estimate the lowest price likely to win and bid that, rather than bidding their full valuation. That means:

• A bid is not a buyer's true value. It is the buyer's guess at the clearing price.
• Floors influence those guesses. Shading algorithms learn from where your auctions clear and where your floors sit.
• A floor set too high does not force buyers to pay more. It often just removes them from the auction.

Why B2B and niche sites need a different approach

High-volume consumer sites can rely on dense auctions, with many bidders per impression, to push prices up. B2B sites often have two or three meaningful bidders per impression, and much of the value comes from a small number of advertisers who know the audience. In that setting:

• Floors matter more, because there is less competition to set a price.
• Mistakes are more expensive, because losing one bidder can halve the auction.
• Direct and deal relationships matter more than in most verticals, because the buyers who value the audience are identifiable.

Step 1: Know where floors are set

Most stacks apply floors in more than one place, and it is common to find them contradicting each other:

• Ad server floors, such as Ad Manager unified pricing rules, which apply to indirect demand including AdX and Open Bidding.
• Wrapper floors, such as those set through Prebid's Price Floors module, which can vary by media type, size, domain, GPT slot or ad unit.
• SSP floors, set in each SSP's own interface.
• Deal floors, set per private marketplace or preferred deal.

Document all of them in one sheet. Where they overlap, the highest effective floor wins, and it may not be the one you think.

Step 2: Segment floors by what actually drives value

A single site-wide floor wastes the most valuable impressions and blocks the rest. Better dimensions for B2B publishers include:

• Content section. A cloud security article and a general industry news brief attract different buyers.
• Ad unit and size. High-viewability in-content units deserve higher floors than below-the-fold units.
• Device. B2B audiences often skew desktop during the workday, and desktop and mobile demand can differ considerably.
• Geography. Where your audience is international, demand varies widely by country.

Step 3: Test with a control group

Floors should always be tested against a control. The Prebid floors module includes a skipRate setting that skips floor enforcement on a set percentage of auctions, and its schema supports running multiple floor models with weights. Use those features to compare revenue per thousand pageviews between floored and unfloored traffic. Look at total revenue, not just CPM: raising CPM while losing 20% of impressions is usually a loss.

Step 4: Move the best buyers into deals

For the handful of advertisers who clearly value your audience, open-market floors are the wrong tool. Offer private marketplace or preferred deals with a fixed or floor price that reflects the audience's value, clear audience definitions, and first-party context such as content categories or company-size segments where you have consented data. Deals give buyers predictability and give you pricing power without guessing at their shading behavior.

Step 5: Keep floors dynamic, but not twitchy

Demand for B2B audiences follows a calendar: fiscal year planning, trade show seasons, product launch cycles, and a strong fourth quarter. Floors should adjust to that. But changing floors every few hours on thin auctions can confuse shading algorithms and add noise. Review floors weekly, adjust based on tested evidence, and leave a record of what changed and why.

Watch the Ad Manager remedies case

In the ongoing US antitrust case over Google's ad tech business, Google proposed in May to stop applying unified pricing rules to open-web display inventory, as part of its suggested remedies. The court has not ruled on remedies, and the trial begins in September. If that change happens, floor management would shift further toward wrappers, SSPs and deals. Publishers with documented, tested floor strategies will adjust more easily.

A floor health checklist

1. List every floor in every system and resolve conflicts.
2. Segment floors by section, unit, device and geography.
3. Test floors against a control, measuring revenue per pageview.
4. Move high-value buyers into deals.
5. Review weekly; change on evidence, not instinct.
6. Pass floor data to your analytics so every rule can be evaluated.

Floors are not a set-and-forget setting, especially for publishers whose audiences are small and valuable. Treated as an ongoing experiment, they are one of the strongest yield levers available. It is the kind of continuous testing a managed team like HBDR runs daily, but the method above works for any ad ops team with good data.