-
Tension: Marketers trust automated buying to place every impression in front of a living, breathing viewer—yet a silent share of spend evaporates the moment the screen goes dark.
-
Noise: Industry chatter frames “ghost impressions” as a technical glitch or fraud problem, masking the broader gap between how CTV is bought and how people actually watch.
-
Direct Message: Until advertisers treat attention as a verified outcome—not an assumed by-product—every programmatic dollar remains half-spent, half-imagined.
Read more about our approach → The Direct Message Methodology
A decade ago, the promise of connected television (CTV) felt almost magical: precise targeting, cinematic screens, and measurable results that linear could only hint at. Fast-forward to today and we know the fine print. In June 2022, GroupM and iSpot.ai uncovered what has since been dubbed the “ghost-impression problem”: bid requests keep firing when many CTV sets are physically off, burning an estimated $1 billion in annual U.S. ad spend.
If that figure sounds dated, consider this: the same pattern persists in 2025, albeit under newer acronyms and tighter privacy rules. Ad verification firms still flag 6-to-9 percent of programmatic CTV impressions as non-viewable because the panel or the set-top stick is alive while the glass is black. For brands now pressured to prove every cent—both for CFOs and for sustainability audits—those unseen messages are more than waste. They are a psychological liability, an uncomfortable hint that the data era is still gambling on faith.
This explainer cuts through the tech speak to reveal why ghost impressions exist, what human assumptions keep them alive, and how reframing “viewability” around verified attention—rather than device status—can finally exorcise the billion-dollar specter.
What it is / how it works
A chain reaction in microseconds
When a viewer launches a streaming app on a Roku stick, that device begins requesting ad slots from exchanges. If the user later switches the TV off with the remote, the stick often keeps drawing power. From the exchange’s point of view, nothing changed; it keeps bidding, winning, and “serving” ads into a digital void. Because CTV lacks the browser-level signals that tell web buyers a tab is closed, the supply chain assumes the set is still on.
Why smart-TV apps fare better
Smart-TV operating systems such as Samsung Tizen and LG webOS manage power states more strictly. When the screen powers down, the OS usually suspends app activity, shutting off ad calls. External dongles and gaming consoles lack that handshake with the panel, so they continue streaming background data—and programmatic bidders oblige.
Measurement blind spots
Traditional verification tags were designed for browsers: an ad calls a JavaScript snippet that checks viewport visibility. On CTV, there’s no equivalent sandbox, so media buyers rely on server-side logs that cannot confirm by a human eye. Until recently, the industry counted an impression if the ad file was delivered, not displayed—language inherited from 1990s web standards.
The deeper tension—attention versus assumption
We buy presence, but we measure possibility
Brands pay for the possibility that someone might see an ad, then reassure themselves with reach curves and modeled frequency. Ghost impressions expose the flaw: possibility is not presence. The deeper tension is emotional: admitting that a budget line might be vapor is frightening, so we default to averages.
“Set it and forget it” meets human habit
Programmatic dashboards seduce us with the illusion of omniscience—sliders, forecasts, instant feedback. Actual viewers behave messily: they fall asleep on the couch, toggle inputs, or yank the power cord. The expectation-reality gap widens every time the ad stack treats a device ID as a proxy for consciousness.
The respect problem
At its heart, advertising is a bid for attention. Serving ads to a blank screen violates an unwritten pact with audiences: I will interrupt only if I believe you’re there.
The more that pact is broken, the more streaming viewers deploy ad blockers, fast-forward hacks, or simply retreat to paid subscriptions. Ghost impressions erode the long-term trust that keeps ad-supported ecosystems viable.
What gets in the way—sources of noise
1) Conventional wisdom: “CTV is inherently measurable”
Industry decks still contrast linear’s panel estimates with CTV’s server logs, implying the latter equals truth. That cliché discourages deeper audits and leaves finance teams blind to off-screen leakage.
2) Fragmented standards
The IAB Tech Lab’s Open Measurement SDK covers mobile and web video but has only recently added CTV guidelines, leaving implementation patchy. Without a universal flag that says device-on = true, each hardware maker invents its own logic.
3) Status anxiety in media buying
When planners chase low CPMs to prove efficiency, they gravitate to open-exchange inventory — precisely where ghost impressions thrive. Pushing for “premium, fully viewable” supply risks higher rates and tougher questions from procurement, so inertia wins.
4) The false fraud narrative
Some vendors label TV-off impressions “fraud,” which frames the issue as malicious. In reality, most leakage is accidental: an OS that never received a power-down signal or an app that buffers for a paused binge. Calling it fraud lets platforms shrug—“we’re not the bad guys”—instead of architecting a fix.
The Direct Message
Attention is an outcome, not an assumption. Until CTV buying verifies that outcome, every campaign includes a silent tax paid to empty rooms.
Integrating this insight—toward attention-verified CTV
1) Contract for power-state signals
Brands wield more leverage than they realize. Make power-state transparency a non-negotiable clause in insertion orders: if a device cannot confirm the screen is on, the impression is non-billable. When enough dollars speak, OEMs will prioritize firmware updates that share on/off status in the bidstream.
2) Prioritize direct supply paths
GroupM’s latest audits show TV-off rates drop below 2 percent when buyers use direct pipes to publisher ad servers versus open exchanges. These curated paths come at a premium, but planners can offset cost by capping frequency or eliminating redundant long-tail inventory that rarely converts.
3) Adopt the “Sustainable Impression” metric
Carbon-accounting frameworks now estimate the energy wasted streaming unviewed video. By tracking grams of CO₂ per viewable impression, brands can tie ghost-impression reduction to ESG goals—a language the C-suite understands. Fewer off-screen ads equal lower emissions and stronger CSR reporting.
4) Experiment with on-device calibration
Several smart-TV makers now expose runtime APIs that ping beacon pixels only when the backlight is active. Piloting these features in measured test markets gives brands hard evidence to defend higher CPMs tied to guaranteed visibility.
5) Educate the finance function
Marketers alone cannot rewire incentives. Share the ghost-impression math with finance partners: a $10 million CTV plan at a 6 percent off-screen rate wastes $600k before creative even loads. When accountants grasp the leakage, procurement is likelier to approve higher-quality, lower-waste buys.
Conclusion
Ghost impressions are not a fleeting glitch; they are a mirror reflecting the gap between how we wish media worked and how devices actually behave.
Each off-screen ad is a missed chance to converse with a real human and an invisible drag on ROI, brand trust, and even planetary resources.
The solution is uncomfortable but clear: move the definition of success from served to seen. That pivot demands collaboration—hardware engineers surfacing power states, exchanges passing the flag, verification firms certifying it, and buyers insisting on the chain of custody.
When attention becomes a verified currency, CTV can fulfill its early promise: cinematic storytelling delivered precisely when a viewer is present to receive it.
Until then, every campaign budget should carry a line item labeled “specter tax.” The only question is how long brands are willing to keep paying.
