Ask a CTV ad ops team how many times the same viewer saw the same insurance commercial last night, and you’ll get a shrug, not a number. That’s the unglamorous truth underneath every panel discussion about “overfrequency” — the industry has a widely acknowledged problem and no shared plumbing to solve it. The latest reminder came via AdExchanger’s Monday roundup, which flagged frequency capping as one of the live issues facing video and CTV sellers right now.
The pitch for fixing it sounds simple: cap how often a single viewer sees the same ad, across every app, device and platform they touch, and everybody wins. Viewers get less fatigue. Advertisers stop wasting spend on the eleventh impression. Publishers get to charge for a cleaner, more valuable inventory pool. The pitch has circulated in ad tech for years, built around vendors like LiveRamp’s identity graph and The Trade Desk’s Unified ID 2.0, both pitched as the connective tissue that lets a frequency cap survive a jump from a smart TV app to a browser tab. What’s changed is the assumption that this infrastructure has matured enough to actually deliver cross-platform capping in production, not just in a case study deck. That assumption deserves more scrutiny than it’s getting.
The Identity Problem Doesn’t Go Away, It Just Moves
Frequency capping inside a single app is trivial — a login-based ID and a counter. Frequency capping across a football game on one streaming app, a news clip on a publisher’s site, and a pre-roll on YouTube is a different animal entirely, because it requires linking three different device or cookie signals to the same actual human being in something close to real time.
That’s identity resolution, and it’s the unsolved problem the industry has circled since cookie deprecation started reshaping the open web. LiveRamp’s RampID and The Trade Desk’s UID2 both attempt probabilistic or deterministic matching across environments, and FreeWheel — Comcast’s ad tech arm, which handles a large share of CTV ad decisioning — has built its own cross-screen measurement layer for exactly this purpose. None of these give a publisher a complete, real-time picture of every impression a given viewer has seen across competitors’ properties, and no publisher wants to hand that visibility to a rival’s ad server anyway. Cross-platform frequency capping doesn’t just need better identity tech. It needs cooperation between parties whose business models depend on not fully sharing what they know about a viewer.
Real-Time Bidding Wasn’t Built for This
Even with usable identity signals, there’s a coordination problem sitting underneath the technical one. Programmatic video runs on real-time bidding, where decisions happen in the milliseconds before an ad serves, auction by auction, platform by platform. A true frequency cap requires that decision to account for what happened in a different auction, on a different platform, seconds or minutes earlier. That means either a shared, cross-platform log of exposure that updates continuously, or new latency injected into every bid request while systems check against it.
Publishers running header bidding setups already manage latency tradeoffs constantly — every additional bidder or verification call slows the page or the stream and risks losing the auction entirely. Layering in a cross-platform frequency check means either accepting slower ad decisioning or trusting a third-party intermediary to make the call, and that intermediary becomes the entity with the fullest view of viewer exposure across the ecosystem. That’s not a publisher. That’s whichever DSP, identity vendor, or supply-side aggregator builds the pipe first — the same dynamic Innovid and Nielsen have each tried to position around with their own cross-screen measurement products, none of which currently function as a neutral, publisher-controlled utility.
Who Actually Captures the Value
Solving overfrequency requires a centralized layer that sees across platforms — which is precisely the kind of infrastructure that concentrates leverage with whoever operates it. If a DSP or an identity vendor builds the cross-platform frequency graph, it’s the one selling advertisers “smarter” reach, and it’s the one with the cleanest data asset to license or resell. The publisher supplying the video inventory gets described as a beneficiary in the sales deck, but the actual product — the deduplicated, cross-platform view of the audience — sits with the vendor whose graph it runs on, not the publisher whose stream generated the impressions in the first place.
Publishers evaluating a frequency-capping vendor partnership should be asking pointed questions before committing engineering time: Who owns the resulting identity graph — LiveRamp, The Trade Desk, FreeWheel, or the publisher? Does the publisher get access to cross-platform exposure data on its own audience, or just a promise that ads will perform better? What’s the latency cost to the publisher’s own auctions, and who absorbs it?
What Publishers Can Actually Do Now
None of this means frequency management is a lost cause at the publisher level. Capping frequency within owned and operated properties — across a publisher’s own app, site, and connected-TV channel — is achievable today with existing ad server tools like Google Ad Manager or FreeWheel’s own server-side stack, and doesn’t require solving identity resolution at internet scale. That’s a smaller, controllable problem, and it’s the one most video publishers should be optimizing before betting infrastructure budget on a cross-platform fix that’s still, largely, a vendor promise.
The bigger fix — the one that actually reduces overfrequency across an advertiser’s entire campaign, not just one publisher’s slice of it — requires industry-wide plumbing that doesn’t exist yet, run by parties whose incentives don’t obviously align with the publishers supplying the inventory. Until that changes, the realistic move for video publishers is to fix what’s inside their own walls and treat every cross-platform frequency-capping pitch as a question about who ends up owning the data, not just whose viewers stop seeing the same ad twelve times.
