Every TV budget conversation in 2026 eventually turns into a debate over how much of a single TV investment should sit in CTV versus linear. The data says the right answer is smaller than most plans assume.
What the efficiency data actually shows
The Efficiency Curve study, conducted by iSpot and LG Ad Solutions, found that most brands unlock meaningful gains by reallocating just 6.3% of total TV impressions from linear to CTV, producing more balanced frequency, broader reach, and stronger conversion. Notably, the top-performing brands in the study aren’t the ones who went furthest toward streaming: they allocate an average of 23.7% of TV impressions to CTV, compared to 17.4% for other brands, meaning even the best-optimized plans still keep the clear majority of impressions in linear.
Linear still carries the ad-supported audience
That restraint lines up with where viewers actually are. Nielsen’s Q1 2026 Ad Supported Gauge found broadcast and cable together held a combined 53.4% share of ad-supported TV viewing, even as streaming set a record high of 46.6%. Linear remains the majority environment for reaching audiences who see ads at all, which is the audience that matters for a media budget.
Why the market average isn’t your answer
An industry-wide spending trend is an average across thousands of advertisers, categories, and inventory strategies. It reflects what the market did in aggregate, not what any single brand’s audience responds to. A packaged goods brand chasing broad reach and a subscription software brand running high-frequency retargeting are not sitting on the same efficiency curve, even if they’re reading the same forecast. Matching the market’s pace assumes every brand sits at the same point on that curve, when the data shows most hit diminishing returns well before the industry average gets there.
At Media Manager, we run incrementality testing to find the actual optimal split for each client’s audience and category, rather than matching whatever the broader market is doing that quarter. Most of the time, that split still leans heavily linear, and the data backs it up.
