Alex Yip, Director of Product Strategy, AppsFlyer
While performance teams often find good results in initial tests with CTV, things often break down when they start scaling campaigns up to serious budgets. The key issue here, says Alex Yip, Director of Product Strategy at AppsFlyer, is that the subset of CTV inventory which delivers reliable outcomes at higher budgets is small. So how can performance marketers win in CTV?
Performance buyers keep approaching CTV with a familiar mental model: if outcomes look good in tests, budgets should scale. In practice, CTV is increasingly failing that simple test. That’s not because inventory is scarce, it’s because performance-eligible inventory is. There may be plenty of nominal “available” inventory. However, the subset that can scale while still holding outcomes is much smaller – and it gets smaller again once you filter out fraud/IVT risk, resold impressions, and indirect supply paths.
That’s the bottleneck: demand is rising faster than the number of clean, outcome-reliable routes to market.
Why “unfilled CTV” isn’t a proxy for performance scalability
Most performance teams are now comfortable running initial tests on CTV. The recurring failure point arrives at the exact moment performance marketers care about most: scaling from “this works” to “this works at $100K/day.” This isn’t an argument against performance CTV. It’s an argument that “inventory volume” is the wrong scaling metric. The constraint is the much smaller subset of supply that stays outcome-reliable as spend rises, and that subset shrinks again once you filter for fraud risk, resold impressions, and indirect supply paths.
Early tests can show strong results, then hit a ceiling because the supply that behaves predictably under higher spend is constrained. As budgets push upward, the channel also attracts more opportunism. More money flowing into performance CTV tends to correlate with more fraud attempts and more “creative” supply packaging. It also tends to mean more variability introduced by intermediated paths.
And even well-intentioned supply strategies create confusion: “direct,” “reseller,” and even what counts as IVT are not consistently defined across the ecosystem, so buyers end up debating semantics while trying to protect outcomes. Meanwhile, the gap between “all available CTV” and “CTV you can rely on for performance” keeps widening.
Test in the pipes, then go direct
A pattern is already emerging: advertisers use DSPs as a sampling layer, then migrate spend to OEMs and streamers once they’ve proven a supply source can drive outcomes. This shift is partly about trust and signal quality (direct platforms often have stronger deterministic signals), but it’s also cold economics: intermediated paths frequently add a meaningful markup, and performance buyers increasingly view that as an avoidable tax once a tactic is validated.
Yes, going direct introduces operational overhead and fragmentation, but many buyers are doing the math and deciding the complexity is still cheaper than paying the toll. The key accelerant is that supply-side tooling is no longer primitive. Many sellers can stand up more direct, programmatic-friendly capabilities “off the shelf” than they could in earlier eras, lowering the barrier to direct paths at scale.
The next constraint will be measurement and attribution. As performance CTV matures, sophisticated advertisers are moving beyond last-click and platform-reported outcomes toward incrementality requirements. The moment you demand incrementality (in parallel with attribution), a lot of “scalable” supply stops looking scalable, because it can’t consistently clear a higher bar of causal impact.
Advertisers should treat CTV exposure as the top of a retargeting sequence, then re-engage on other screens before a user ever hits a site. That strategy is powerful, and it reinforces the bottleneck because it depends on clean connectivity, consistent signals, and a lot of integration work across platforms. The industry is already feeling the weight of “too many pixels, too many APIs, too many pipes,” and 2026 is shaping up as the year those plumbing problems become impossible to ignore.
The winners in 2026 will define “performance-eligible CTV supply” and operationalise it. They should:
- Separate exploration budgets (DSP-heavy) from scaling budgets (cleaner, more direct paths)
- Adopt incrementality as a gate for scaling, not a report after the fact
- Invest early in identity/signal plumbing and cross-screen retargeting design, because that’s where performance CTV is actually headed
Performance CTV is running out of trustworthy, outcome-sustaining routes to market. The buyers who treat that as a supply-chain problem in 2026 will be the ones who keep scaling and building on the benefits of this unique performance channel.
Follow VideoWeek on LinkedIn.
