Turning premium CTV into performance revenue with EX.CO’s Lior Yossef
ROB BEELER, BEELER.TECH + LIOR YOSSEF, DIRECTOR OF PRODUCT AT EX.CO
CTV has long been positioned as the crown jewel of digital media – brand-safe, high-attention, living room inventory that commands premium budgets. Yet behind the promise of premium CPMs, many media owners are discovering a tougher reality: yield volatility, opaque auction dynamics, and monetization stacks that haven’t evolved at the same pace as demand.
In this conversation, Rob Beeler joins Lior Yossef, Director of Product at EX.CO, to unpack where CTV monetization is falling short, and what media owners should be doing differently. Together, they explore the disconnect between “premium” perception and practical yield mechanics, the risks of biased optimization, and why machine learning must move beyond buzzword status to real-time execution.
From demand orchestration to transparency in reporting, this discussion zeroes in on how CTV stacks can shift from reactive revenue management to precision-driven performance.
Rob: CTV is often described as “premium inventory,” but many owners still struggle to maximize yield. Where do you see the biggest disconnect between perception and reality in CTV monetization?
Lior: The biggest disconnect is assuming that “premium” automatically means “well monetized.” CTV inventory is premium from a viewer and brand perspective, but the monetization mechanics are still immature in many stacks. Too often, media owners rely on static assumptions – such as fixed floors, preferred demand paths, or legacy deals – without continuously validating whether those choices reflect how demand is actually behaving in real time. Premium content doesn’t protect you from inefficiencies; it just makes those inefficiencies more expensive.
Rob: Unlike web or mobile, CTV has fewer impressions and higher stakes per auction. How does that change the way media owners should think about yield optimization in CTV environments?
Lior: In CTV, every impression carries more weight, so optimization has to be about precision, not volume. You don’t have the luxury of “learning later” or averaging performance across millions of impressions. That means media owners should focus less on blunt tactics like aggressive floors and more on understanding how each auction decision impacts long-term yield, fill stability, and buyer trust. In CTV, bad decisions compound quickly – and good ones do too.
Rob: Many CTV stacks rely on multiple demand partners with different behaviors and incentives. What typically breaks first when those pipes aren’t truly working together?
Lior: Many of the programmatic pipes in CTV are fundamentally broken – not because of any single partner, but because they were never designed to work together in real time. Each demand source operates with its own incentives, pacing logic, and bidding behavior, yet most stacks treat them as interchangeable.
What breaks first is auction coherence: signals conflict, competition becomes uneven, and pricing decisions are made in silos rather than as part of a unified system. Without coordination across demand, publishers lose a clear view of true bid pressure, floors become blunt instruments, and yield becomes volatile. Until those pipes are orchestrated rather than simply connected, CTV monetization will continue to underperform its potential.
Rob: You’ve talked about demand-agnostic optimization. For CTV media owners, what’s the real risk of relying on optimization biased toward a single demand source?
Lior: The risk is mistaking convenience for performance. Optimization that favors a single demand source may look efficient in the short term, but it quietly limits competition and masks missed opportunities. Over time, that bias can train the system to under-value impressions, weaken negotiating leverage, and make revenue overly dependent on one buyer’s strategy. True yield optimization should serve the inventory, not the demand source with the loudest voice.
Rob: Machine learning gets thrown around a lot in adtech. In CTV specifically, which decisions actually benefit from real-time machine learning, and which ones don’t?
Lior: Machine learning is most effective in CTV decisions that change auction by auction, such as dynamic floor pricing, demand selection, and predicting win probability based on live market conditions. CTV performance depends on many contextual and user-level signals that directly impact performance, making manual decision-making extremely complex and nearly impossible to manage at scale. Because CTV has fewer impressions and higher value per opportunity, historical averages and static rules break down quickly.
At EX.CO, our machine-learning yield engine analyzes billions of signals across supply, demand, and auction dynamics to forecast performance in real time and continuously adjust pricing and demand paths at the impression level. This enables optimization based on actual buyer behavior as it’s happening, rather than after the fact.
Strategic decisions like partner selection, content packaging, or deal structure still require human judgment and business context, but machine learning is essential for executing those strategies efficiently at scale.
Rob: Transparency is harder in CTV than in other channels. What should CTV media owners insist on seeing in reporting if they want to understand what’s truly driving yield?
Lior: They should insist on reporting that shows why an auction resolved the way it did, not just the final CPM. That means visibility into response rate, win rate, and request RPM. Without that context, it’s impossible to tell whether revenue is truly being maximized or simply accepted.
Rob: As CTV inventory expands across apps, FAST channels, and new screens, what operational mistakes do you see media owners repeating as they scale?
Lior: The most common mistake is scaling complexity without scaling control. Media owners often add new endpoints, channels, and partners while keeping the same monetization logic underneath. That leads to fragmented decision-making, inconsistent pricing, and operational blind spots. Scaling CTV successfully requires centralized intelligence – even if distribution is decentralized.
Rob: If a CTV media owner wanted to pressure-test whether their stack is working for them or against them, what’s the single question they should be asking right now?
Lior: They should ask: “If demand behavior changed tomorrow, would my stack adapt automatically – or would revenue lag while we caught up?” That question cuts straight to whether optimization is proactive or reactive. In CTV, the difference between the two is often the difference between protecting yield and slowly giving it away.
Ready to turn premium CTV into performance revenue?
CTV’s promise isn’t going away – but neither are the structural inefficiencies that limit performance. As inventory scales and buyer behavior evolves, media owners need more than premium positioning; they need precision orchestration, demand-agnostic optimization, and real-time intelligence that adapts as quickly as the market does. The stacks that win won’t be the loudest or the most complex – they’ll be the most coordinated.
If you’re ready to rethink how your CTV monetization strategy performs under pressure, explore EX.CO’s CTV offering at: ex.co/ad-server.
👉Want to stay in the loop with what’s happening in the Beeler.Tech community? Subscribe to our newsletter. If you’re interested in attending or sponsoring a future event, you can explore our upcoming events here.
This is content created in paid partnership with EX.CO. We only feature partners who we believe bring real value to the publisher community.