Setting RTB floor prices by vertical: a 2026 playbook

Setting RTB floor prices is one of the few places where being wrong in either direction costs real money the same day. Too high, and your fill rate collapses — buyers route to competing networks and your publisher relationships erode. Too low, and arbitrage buyers strip your inventory at a profit while you subsidize their margin. Here's the framework we use across pay-per-call networks running on Pulse RTB.

Vertical benchmarks for 2026 (US markets)

These ranges reflect production data from networks running on Pulse RTB across the three highest-volume US pay-per-call verticals. They assume standard quality thresholds: 90+ second duration, verified US origination, no repeat-dialer flag, first-time caller to the buyer.

The fill rate vs. quality tradeoff

Most operators think of floor pricing as a one-dimensional lever — raise it to make more money per call, lower it to fill more calls. That framing misses the more important variable: quality-adjusted fill rate.

A network with an 80% fill rate at $45 floor and 8% fraud rate is delivering worse economics than one with a 60% fill rate at $65 floor and 2% fraud rate. The math: at $45 with 8% fraud, effective revenue per impression is $45 × 0.80 × 0.92 = $33.12. At $65 with 2% fraud: $65 × 0.60 × 0.98 = $38.22. Higher floor, lower fill — better outcomes.

The mechanism that breaks this analysis: buyer concentration. If 60% of your impressions route to three buyers, and those buyers have specific quality thresholds that your floor pricing doesn't capture, you'll fill less than you'd expect even when you price correctly. Map your buyer quality tiers before setting floors.

Dynamic flooring: the four adjustment signals

Static floor prices are a starting point. What moves revenue is dynamic adjustment across four dimensions:

1. Time of day. Buyer demand peaks 8am–11am and 6pm–8pm local time for most US verticals. Floors can support 12–18% premiums during these windows without meaningfully impacting fill rate, because buyers are competing for scarce inventory.
2. DMA quality tier. Buyers in insurance, mortgage and solar consistently pay premiums for calls from top-25 DMAs. A call from Dallas or Phoenix is worth 20–35% more than the same caller profile from a Tier-3 DMA. Floor pricing should reflect this.
3. Buyer demand signal. When win-rate on a given buyer drops below 25% over a 4-hour rolling window, reduce the floor 8–12% to recover impressions. When win-rate exceeds 70%, raise it. This is standard auction theory applied to your specific demand pool.
4. Publisher quality signal. Publishers with strong track records (low fraud rate, high conversion rate for buyers) should get preferred routing with a tighter floor — you want to maximize volume from good publishers. Publishers with borderline quality metrics should face higher floors or minimum quality scores, not just lower payouts.

A/B testing floor prices without tanking your revenue

You can't test floor prices the way you'd A/B test a landing page — you're not splitting user traffic, you're splitting auction inventory, which has memory effects (buyer bid strategies adapt). Here's the approach that minimizes risk:

The biggest mistake we see in floor price testing: operators lower floors to "improve fill rate" after a slow week, without checking whether the slow week was demand-side or supply-side. Buyer-side softness doesn't respond to floor reductions — you need to check your buyer win-rate and bid depth before touching floors.