232 Publisher Sites. 7 Findings. One Managed Stack.

This is the finding that stopped us cold.

Across all 232 publisher sites in this group, zero are running price floors. Not one. In 2026, when floor pricing is table stakes for any publisher serious about yield, an entire managed portfolio is going into auctions with no minimum price protection whatsoever.

What does this mean in practice? Every low-quality, sub-$0.10 bid from a demand partner clears just as easily as a premium bid. There is no mechanism protecting publishers from race-to-the-bottom CPMs. The auction has no floor, so the floor is effectively zero.

Floor pricing is not experimental. Prebid's floor module is mature, widely adopted, and directly correlated with yield improvement. Its complete absence across a managed portfolio this size raises a serious question: who is this setup optimized for?

Prebid.js is actively maintained with regular performance improvements, new bidder support, and security patches. Across this publisher group, here's what we actually observed running in production:

The 65 sites where no version was detected includes: BBC, Bloomberg, Washington Post, Vogue, The New Yorker, Wired, Vanity Fair, AOL, and others. Whether these are non-standard builds, fully wrapped implementations, or something else entirely — the opacity itself is a problem for publishers expecting transparency into their own configuration.

Publishers on v8 are missing two full major versions of improvements. In a managed program, version fragmentation at this scale suggests there is no centralized update cadence.

The highest-CPM demand partners in this dataset are also among the slowest to respond. Kargo averages $1.09 CPM at 328ms. Triplelift averages $0.91 CPM at 422ms. OpenX averages $0.86 CPM at 554ms.

Now look at what timeout settings we observed across major publishers in this portfolio:

A 1,000ms timeout in a stack where premium bidders respond between 400–750ms means you are systematically excluding your highest-paying demand on some of your most valuable inventory. Every impression on Wired.com or Vogue.com that times out before a premium bid arrives is revenue that never gets captured.

The data suggests an optimal timeout range of 2,000–3,000ms for this bidder mix. 20 total sites in this portfolio are running below 1,500ms.

As the industry moves beyond third-party cookies, identity resolution has become a fundamental requirement for maintaining addressability with premium buyers. Our scan found:

Sites with zero identity infrastructure are invisible to demand partners that rely on addressable targeting. When premium buyers cannot match a user to their data, they either don't bid or bid at a significant discount.

Additionally, zero sites across the entire 232-site portfolio have any RTD (Real-Time Data) modules deployed. RTD modules enable audience enrichment and contextual signals that improve bid density and CPM. Their complete absence across the entire portfolio is another systematically missed lever for yield improvement.

When we look at bidder win rates across the group, a pattern emerges that publishers should find troubling: multiple demand partners are generating significant bid activity while contributing zero auction wins.

These partners are consuming auction capacity and adding latency overhead without generating a single dollar of revenue in our observation window. In a well-managed stack, consistently underperforming partners get audited and rotated. In an unmonitored one, they just sit there — indefinitely.

Managed programs often route some demand through server-to-server (S2S) configurations to reduce client-side latency. The tradeoff is that S2S bids frequently come in lower than their client-side equivalents due to reduced signal fidelity. In this portfolio, we can observe both paths head-to-head:

The S2S bids are coming in at a fraction of the client-side equivalent CPMs and winning essentially nothing. Rubicon via S2S bids at $0.17 vs $0.58 client-side. Pubmatic via S2S bids at $0.13 vs $0.43 client-side.

If publishers understood this dynamic, they would have serious questions about whether the S2S routing is benefiting them — or the managed provider's infrastructure.

Kargo stands out as the single best-performing demand partner in this entire dataset by average CPM. It checks every box: high CPM, healthy win rate, fast response time.

Kargo averages $1.09 CPM, wins 14.3% of its bids, and responds in 328ms. It was observed bidding on only 42 impression opportunities across 232 sites.

Compare that to Index Exchange (402 bids), Criteo (373 bids), or MediaNet (339 bids) — all bidding at less than half the CPM. Kargo is dramatically underdeployed relative to its performance profile. A managed partner with real visibility into bidder performance data would be expanding Kargo's footprint across the portfolio. Instead it appears to be a minor participant while lower-performing bidders dominate bid volume.