CTO Series 1: Retail Media in Volatile Markets: Why Static Floors Break

‍By Francisco Larrain, Co-Founder & CTO, Topsort

‍

Retail media looks simple from the outside. An ad request comes in. Advertisers bid. A winner is chosen.

Clean. Fast. Simple.

But in volatile markets, that simplicity collapses fast. What changes is not the mechanics of the auction, it's the environment around it. Promo cycles shift demand overnight. Bid density fluctuates week to week. New sellers enter and exit. Seasonality changes conversion behavior. Shopper intent moves faster than floor rules.

And that's where static floor pricing fails.

The Problem With Static Floors

A static floor assumes the bid landscape is stable. But retail demand is not stable.

If your floor is too low during high demand, then you leak yield. If your floor is too high during low demand, then you kill fill and hurt advertiser confidence.

In volatile environments, static floors create oscillation. Revenue spikes, then crashes, then ad ops manually adjusts, then overcorrects. The retailers we see growing ad revenue consistently month over month are almost never the ones manually tuning floors.

This is not scalable retail media infrastructure. It's manual firefighting.

Retail Media Is Not Publisher Media

Most ad server technology was built for publishers. Publishers monetize attention. Retailers monetize transactions. That difference matters more than people admit.

In retail media:

• Product availability changes placement value in real time
• Conversion probability varies by placement and moment
• Margin and category dynamics affect what a slot is worth
• Shopper experience has an opportunity cost that doesn't show up in traditional media math

You cannot treat floors as simple CPM guardrails. They are part of an economic system, and if you're using publisher-era logic to manage them, you're working from the wrong model.

What Modern RMNs Do Instead

Leading retail media networks have moved to dynamic floor strategies. Not because it sounds sophisticated. Because volatility eventually forces the issue.

A dynamic approach:

• Responds to real-time demand signals
• Adjusts to bid density changes as they happen
• Protects fill without sacrificing yield
• Reduces reliance on manual tuning

When bid density is up, the floor moves up. When demand is thin, it comes down. The system reads the same signals the auction is already processing and makes the call in real time, not three days later when someone notices a dashboard anomaly.

One thing worth being clear about: this is not a mechanism to squeeze advertisers. The goal is to maximize value per opportunity while protecting ecosystem health: advertiser ROI, fill rate, shopper experience, and revenue, all at once. That requires infrastructure, not spreadsheets.

The CTO View

At scale, floor management becomes a control problem.

You're balancing revenue, fill rate, advertiser ROI stability, and shopper experience simultaneously, in real time. These forces pull against each other in ways that aren't always obvious, and if they're manually coordinated, the system will eventually break under growth. Not because the team isn't good enough. Because the problem scales faster than any team can.

Retail Media 3.0 requires monetization logic embedded into the auction layer itself. Not bolted on outside of it. The auction mechanics are table stakes at this point. The yield intelligence underneath—that's where most networks are still flying blind.

Francisco Larrain is Co-Founder & CTO of Topsort.

Curious how Topsort approaches monetization infrastructure? Book a demo.