Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Floor Optimization

A floor price protects a publisher from selling attention below its worth — set too low, second-price clearing grinds toward zero in thin competition; set too high, fill rate — the share of slots that actually serve an ad — collapses. Publishers shouldn’t have to guess this number, so Promovolve measures it.

Sweep, don’t learn

The FloorSweepOptimizer is deliberately not reinforcement learning — an earlier RL approach (value estimates over discretized floor levels) was built, evaluated, and dropped: with pacing and Thompson Sampling already adapting around every floor change, credit assignment was hopeless, and the agent mostly learned noise. What replaced it is controlled measurement:

  1. Sweep. Generate candidate floors across the plausible range — bounded below by observed rejected bids, above by 99% of the second-highest approved bid. Not the top bid: any floor between the second and top bid manufactures a monopoly (only the top bidder clears), and since each campaign fills at most one slot per page, pricing out the rest of the field forfeits their fill for a few percent more on one impression. A floor may price out the bottom of the field, never the top two; extraction above the second bid is second-price clearing’s job, not the reserve’s. Hold each candidate for a fixed number of auction ticks, measuring served revenue — actual post-pacing, post-selection earnings, not theoretical clearing prices. Revenue is the only honest objective; anything upstream of it can be gamed by the very systems the floor interacts with.
  2. Exploit. Take the argmax — the floor that earned the most — and hold it for a longer exploitation period. Ties within tolerance resolve to the higher floor — the same revenue on fewer impressions, and more robust if the dominant bidder’s value drifts up. The optimizer only fails open to the lowest floor when evidence is missing, and a minimum-impressions guard keeps a lucky low-traffic candidate from winning on a handful of data points.
  3. Repeat. Markets drift; the cycle re-runs continuously.

Per-category floors

A single site-wide floor has a monopoly pathology: one rich category (say, Finance demand at $12) sets a floor that locks out every other category’s demand entirely. Floors therefore run per content category, each category sweeping independently; categories without enough data fall back to the site-level floor. The pathological case that motivated this — a lone high bidder pushing the site floor above everyone else — now prices one category at $12 while Travel still clears at $3.

Two guardrails matter more than the sweep itself:

  • Only approved demand teaches floors. Pending creatives bid (to reach the approval queue) but are invisible to the optimizer — otherwise an unapproved, possibly rejected campaign could inflate a floor that outlives it.

  • A lone bidder pegs the floor to 99% of its bid. With one approved bidder there is no second price and nothing to sweep; the floor snaps to just under the bid — the 1% headroom guarantees the bidder always clears its own floor — and instantly re-derives when the field changes (validated live in both directions, including pausing the top bidder mid-flight). And when there is no demand that can actually serve — zero approved bidders, or bidders whose every bid the floor has priced out — the floor collapses to the minimum immediately: demand that cannot serve is not demand, and a floor with nobody to price against is pure fill-rate damage.

  • Slot quality can only discount a floor, never raise it. The floor a bid must actually beat is the category floor scaled by the slot’s engagement prior, clamped at 1.0× — a weak slot may price below the category floor to attract fill, but a premium slot never surcharges above it. The invariant this preserves is the one every guardrail here serves: the floor a bidder faces is never derived from their own bids plus a markup. Each rule alone looked safe with a multiplier above it; composed, they once strangled a live category — so the invariant is now enforced at the layer auctions actually consume.

One honest caveat: in a perfectly homogeneous market — every bidder at the same CPM — every floor below the common bid earns identical revenue, and the optimizer settles anywhere on that plateau. That is correct behavior, and a reminder of what this component is: not intelligence, just a well-designed experiment that never stops running.