The most expensive problems in programmatic CTV are not the dramatic ones — fraud, outages, disputed measurement. They are the quiet ones, sitting in the plumbing between a publisher’s ad server and an advertiser’s bidder. That is the argument Kieran Greene, founder and CEO of sell-side ad tech firm Shinka, made to VideoWeek on 19 August: the process that shepherds a bid request from the sell side to the buy side routinely operates so inefficiently that publishers never see the full value of their inventory. Publishers typically see a 20 to 40 percent revenue uplift once those bid-path inefficiencies are fixed, according to Greene. That is a vendor’s number — but it is a vendor pointing at leaks any operator can go and check.
Greene says he first saw the problem from inside Google, where the tying of AdX to Ad Manager shaped which demand could compete for an impression — an architecture publishers will recognise from a decade of open-web auction dynamics, now echoed in vertically integrated CTV stacks that translate bid requests diligently for their own demand path and less so for everyone else’s. The leaks he catalogues are unglamorous and specific. Bid requests go out malformed, or missing the datapoints that decide whether an advertiser bids at all: show name, season, episode, ad break position. Geographic signals arrive half-built — Greene’s example is receiving a ZIP code with no city or country attached, leaving the bidder to guess at what should be a routine lookup. And because every DSP and SSP wants the request formatted its own way, translation quality varies by partner; where the work is skipped, demand simply fails to show up. Fewer bidders seeing a well-described impression means lower bid density, less competition, and CPMs that settle below what the inventory would clear in a fair contest.
The mispricing extends beyond the living room. In digital out-of-home, which increasingly trades through the same programmatic pipes, audience multipliers are routinely lost in translation. “An ad seen by fifty people in a bar or forecourt is often traded as if it were seen by one person on their sofa,” as Greene puts it — his firm works with fuel-station network GSTV, where the gap between actual and traded audience is the business model’s core problem.
01The leak is structural, not episodic
What makes this argument worth a publisher’s attention is that none of these failures announce themselves. A malformed bid request does not error out; it just attracts fewer and lower bids. A missing episode-level signal does not break the ad break; it merely converts premium, content-targeted demand into run-of-network pricing. The revenue that should have arrived never appears on any report as a loss, because no system logs the bids that were never made. This is the same class of problem the open web spent a decade learning to audit — and CTV, having inherited programmatic’s pipes, has inherited its silent failure modes, with the added complication that CTV’s most valuable signals are content metadata that many broadcaster stacks still treat as internal information rather than as pricing input.
Greene’s prescription is, unsurprisingly, the shape of his company: an agnostic mediation layer that prepares and enriches requests properly, does the translation work equally for every demand partner, charges flat CPM-based fees rather than a revenue share, and carries SOC 2 Type II certification. Publishers do not need to buy that particular answer to act on the diagnosis. Every leak on the list can be audited with your existing partners, and most of the fixes are configuration and contract questions rather than replatforming decisions.
02Why this matters for publishers
| The loss is invisible by design | Missing signals and skipped translation suppress bids that are never logged anywhere, so standard yield reporting shows a healthy auction while underpricing every impression in it. |
|---|---|
| Content metadata is pricing power | Show, season, episode and pod-position signals are what let advertisers pay premium prices for premium context. Withholding them — deliberately or through pipeline neglect — converts your best inventory to average pricing. |
| Integrated stacks have structural incentives | A sales house whose ad server, SSP and demand arm share an owner has reasons to route value inward; the diligence a publisher owes its own P&L is knowing exactly how every partner translates and distributes its requests. |
| The claimed prize is large even if discounted | A vendor-quoted 20 to 40 percent uplift deserves scepticism, but even a fraction of that range, on CTV CPMs, funds the audit that verifies it many times over. |
Programmatic CTV's yield problem, on this account, is not that demand is absent — it is that the pipes quietly waste it, one malformed request and one missing signal at a time.
03What publishers should do
04The bottom line
Programmatic CTV’s yield problem, on this account, is not that demand is absent — it is that the pipes quietly waste it, one malformed request and one missing signal at a time. Greene is a vendor with a mediation layer to sell, and his 20 to 40 percent figure should be treated as a claim to test rather than a fact to bank. But the leaks he names are real, checkable and largely inside a publisher’s own control: know what your bid requests actually say, make signal completeness a contractual expectation, and measure competition per path instead of trusting the topline. In a market where every negotiation is fought over single CPM points, revenue recovered from your own plumbing is the cheapest money in the building.