Are ad budgets keeping up with the shift from TV to streaming?

There is a well-worn principle in media that reach without relevance is just noise. For decades, the advertising industry built its planning architecture around this tension: television provided reach but couldn't always identify the specific audience. Digital promised to fix that. It handed advertisers an identity layer, a targeting stack and a click to prove it worked. The irony is that in chasing the individual, digital lost something television had always understood instinctively: context shapes attention, and attention shapes outcomes.

That trade-off is now at the centre of one of the most consequential debates in modern media buying. Connected television, the broad category covering smart TVs, streaming devices and internet-delivered content on the big screen, has spent the better part of a decade promising to be the bridge between television's emotional gravity and digital's precision. But a mounting body of data suggests that bridge has a structural gap, and closing it may require the industry to rethink what "data-driven" actually means in a streaming-first world.

The question is no longer simply whether CTV can scale. It already has. The question is whether the infrastructure around it such as the signals, the metadata, the content layer, is sophisticated enough to justify the budgets advertisers are finally ready to commit. The answer, increasingly, is that it is not quite there yet. But it is getting there fast, and what comes next could reshape how media is planned, bought and measured across the entire advertising ecosystem.

Read On: Streaming paradox: Is India’s CTV boom running into a precision problem?

Russhabh R. Thakkar, Founder and CEO of Frodoh, points to what he sees as an oversimplification at the heart of current CTV benchmarking. Completion rates on CTV hover around 95 percent, partly because the environment is largely non-skippable. That number gets cited frequently as evidence of CTV's advertising effectiveness. Thakkar is sceptical. "That alone doesn't mean the advertising worked. What matters is the quality of attention. A family watching content together on a large screen in a living room is a very different engagement environment compared to passive mobile consumption," he argues.

His point lands with particular force given the household dynamics of CTV. Unlike a smartphone, which is an inherently personal device, a connected television is often a shared screen. This creates a co-viewing environment that amplifies reach in ways that standard impression counting does not capture. But it also means that a single impression could be seen by multiple viewers, or by none, if the television is on in the background while the household is occupied elsewhere. The metric of average frequency per user, borrowed wholesale from digital, does not map cleanly onto this environment.

Somendu Singh, Chief Contributor at CTV Scale, identifies this as a structural category error. "The audience duplication we see in the data exists because we are looking at CTV through a digital lens. CTV is a household device with multiple viewers. But since the ad delivery mechanism is digital, the industry treats it as an extension of digital media, assuming one device equals one user," he notes. The implications for planning are significant. Frequency caps, reach curves and budget allocation models that were designed for one-to-one digital environments may be actively distorting how CTV is bought and evaluated.

This is not an argument against CTV. It is an argument for building the right measurement architecture around it. Thakkar puts it plainly: "Planners should stop treating completion rate as the final KPI and start looking at things like attention, brand recall, incremental reach, engagement across devices and actual business outcomes. Campaigns can't just be measured after they end anymore. Brands want live visibility into performance and the ability to optimise while campaigns are active."

The Data Gap That Budget Confidence Cannot Bridge

According to a new study titled "TV Audiences Have Shifted. Ad Dollars Have Not," published by Gracenote, the content intelligence business unit of Nielsen, 89 percent of media planners anticipate shifting more budget from linear TV to CTV over the next 12 to 24 months. The direction of travel is unambiguous. Yet the same report reveals that 47 percent of those planners cite limited show-level or content-level data as a primary barrier to actually accelerating that shift. Intent is there. Confidence is not.

Ryan Moore, Chief Business Officer at Gracenote, frames the issue in terms that every media buyer will recognise. "Buyers aren't asking for more complexity — they're asking for the same transparency they've relied on for decades in linear TV," he says. "Bringing show-level visibility to CTV gives the channel a clearer path to bigger budgets, not just from linear but across the digital video ecosystem. When buyers have the insight to validate placement quality and prove impact, CTV becomes more accountable and competitive."

Read On: Beyond Metros: Are Tier 2 & 3 markets truly powering India’s CTV boom?

The gap is not merely a planning inconvenience. Among programmatic traders who actively transact on CTV inventory in real time, the issue is even more acute. The Gracenote study found that 100 percent of programmatic traders describe show-level transparency as very or extremely important for ensuring brand safety and inventory quality. And 95 percent agree that the absence of show-level signals actively prevents them from advocating for larger CTV budgets during planning conversations. This is not a niche technical complaint. It is a systemic confidence problem that is suppressing capital allocation in what is otherwise the fastest-growing segment of the video advertising market.

What Gracenote defines as the CTV Data Gap is essentially the missing content layer behind an ad impression. While audience targeting in CTV has become increasingly sophisticated allowing advertisers to reach a 25-to-34-year-old sports fan who earns above a certain income threshold—the broader market still lacks a standardised view of what that fan was actually watching when the ad appeared. Without that content signal, buyers lose confidence at the most critical stages of a campaign cycle, from planning to reporting. And without confidence, budgets do not move.

The scale of the opportunity that content intelligence could unlock is significant. The Gracenote study found that 80 percent of traders would shift budget from audience-targeted to contextually targeted CTV if actionable content signals were available. Sixty-five percent said they would also consider reallocating spend from programmatic video, and 63 percent from display advertising. This is not a conversation about CTV merely competing with linear television. It is a conversation about CTV competing for a far larger share of the total digital advertising mix.

From Brand Safety Net to Strategic Growth Engine

For most of the past decade, contextual targeting in digital media was treated as a risk management tool. It was a way to ensure that a bank's advertisement did not appear next to news about a financial scandal, or that a children's brand did not surface alongside adult content. It was the guardrail, not the engine. That positioning is changing rapidly, and nowhere is the shift more visible than in CTV.

The evolution is being driven by a fundamental change in what contextual data can now do. In its earlier iteration, content-level targeting relied on blunt instruments, like, keyword filters, category exclusions, surface-level genre tags. What is emerging now is a far more granular infrastructure.

Read On: IPL 2026 - CTV’s premium reset: Why are advertisers still finding value

Gracenote's Content Graph, for instance, provides advertisers with source-validated metadata that goes well beyond genre. It includes programme-level identifiers, content ratings, language signals and unique content IDs that allow buyers to match ad placement against specific shows, not just broad content categories. This level of specificity transforms contextual intelligence from a brand safety filter into a genuine targeting tool — one that can drive audience relevance, improve campaign efficiency and generate post-campaign reporting that buyers actually trust.

A recent activation provides a useful proof of concept. In the first quarter of 2026, a campaign for Dos Equis, executed by dentsu X through Index Exchange's supply-side platform using Gracenote CTV Content Intelligence, was designed to reach college football fans with precision. The result was that 100 percent of impressions were delivered within targeted college football content, including 84 percent during live College Football Playoff games, while CPMs came in 7 percent under benchmark. Reach, relevance and efficiency in a single campaign — the trifecta that has long eluded CTV at scale.

Medhavi Singh, Country Head at Criteo India, sees this development as part of a larger structural shift in how brands think about signals. "The shift is real, but I wouldn't coin it as a return to traditional content-led planning. Rather an evolution that is being driven by a convergence of forces that have fundamentally changed what signals advertisers can rely on," she says. As identity signals continue to fragment across the open web, content context is becoming a more durable proxy for intent — not because it tells you who someone is, but because it tells you what mindset they are in.

Singh's framing is instructive for understanding why contextual intelligence is attracting so much attention at this particular moment. "Context paired with commerce signals tells you something about intent. That combination is what bridges the gap between brand and performance," she explains. "When a brand can reach a high-attention viewer in a relevant content environment and then trace that exposure through to a real commerce outcome like a product page visit, an add-to-cart, a purchase, and that's when CTV stops being a line item in the brand budget and starts competing for performance spend too."

What Comes Next: Intelligence as Infrastructure

The trajectory is clear even if the destination is still being mapped. CTV is transitioning from a channel defined by its inventory to one defined by its intelligence layer. The ability to connect content context, audience signals, attention quality and commerce outcomes in a single, coherent view of campaign performance is the capability that will determine which platforms and which partners attract the largest budget commitments over the next three to five years.

Thakkar frames the competitive landscape with precision. "Over the next few years, the winners in CTV will not just be companies offering inventory. The real advantage will come from intelligence layers, including planning, contextual understanding, optimisation, and measurement. That's where the industry is moving very quickly now."

Somendu adds a note of measured realism. The ecosystem is not fully evolved. In markets like India, where a significant share of the CTV universe is controlled by a small number of walled gardens, data reliability and signal transparency remain genuine challenges for independent measurement agencies. Attribution, in its truest sense — tracing an ad exposure on a connected television to a downstream purchase — is still aspirational for most advertisers in most markets. "It is too early to say that CTV is a full-funnel advertising medium, as the ecosystem is not yet evolved enough to map attribution to CTV advertising," he observes.

But the direction is not in doubt. The structural forces that are driving investment in content intelligence — identity signal deprecation, the demand for brand-safe and measurable environments, the collapse of the linear-digital binary as the dominant planning framework — are not going away. The next divide in media buying may no longer be between television and digital. It may be between media that can demonstrate contextual relevance and accountable outcomes, and media that cannot.

In that world, the intelligence layer is not a feature. It is the product.