The traditional economics of linear TV advertising are built on tension. Publishers want higher CPMs. Advertisers want lower ones. Every rate negotiation is a zero-sum exchange where one side's gain is the other's loss. This dynamic has defined the business for decades, and it's the reason so much of the buying process remains manual, adversarial, and inefficient. But a growing segment of the market, now representing nearly half of all linear TV revenue, operates on an entirely different logic: performance. In performance, the incentive structure inverts. When an ad drives a measurable outcome, the advertiser can afford to pay more because their cost per sale or cost per response goes down. The publisher earns a higher unit rate because the inventory delivered real value. The question is why the infrastructure to optimize for this has taken so long to arrive.
Long Ellis is the Founder and CEO of Denali TV, an optimization platform for performance-based linear television. He's a former television executive with a track record of scaling revenue and a resume that includes senior roles at CBS, Discovery Communications, and Viacom, among others. With a career that spans both the legacy broadcast business and the ad tech ecosystem built around it, he sees the current moment as the inflection point where programmatic logic finally reaches the largest remaining pool of manually transacted TV inventory.
"In the performance world, if you maximize the outcome, both sides win. The advertiser gets a lower cost per sale and the TV network gets a higher unit rate because they are delivering actual value." In his view, the "tech tax" of digital media has made linear TV, with its verified impressions and high working media percentages, the most undervalued performance asset in a marketer’s arsenal.
The core inefficiency Ellis describes is structural. Performance advertisers bid unit rates across broad daypart rotations, but they don't control where within those rotations their ads actually land. The TV network places them wherever it has holes to fill. The result is that an advertiser might land in the two highest-performing hours of a daypart or the two worst, and they have no way to influence the outcome. "Advertisers are kind of rolling the dice," Ellis says. "They could be in a powerful daypart, but if the network is putting them in the hours that underperform, they're going to get negative data. If there are two or three hours within that daypart that perform 50% better than the average and you can target those hours, you're solving that problem. You're reducing waste and making the spend much more predictable."
Platforms like Denali ingest first-party performance data from the buy side and available inventory from the sell side, then algorithmically match ads to the highest-performing available programming. Neither side's data is shared with the other. The platform acts as an agnostic optimization layer that serves both interests simultaneously. "I don't know of another ad tech platform that literally benefits both sides almost equally."
One of Ellis's sharpest arguments is against the industry's over-reliance on audience-first targeting. Lookalike audience strategies, which dominate connected TV and are increasingly applied to linear, ignore the performance impact of programming context, time of day, and content relevance. "The marketplace is a little confused right now. A lot of advertisers and publishers are trying to target lookalike audiences who have purchased a certain product, but when you do that and you're just targeting an audience, you're neglecting the fact that an ad can perform dramatically better in certain programs, at certain times of day, in certain genres."
He points to research showing that the same ad can perform measurably differently across episodes of the same show, depending on the content of that episode. A breakfast product performs better in morning programming before work. A product aimed at white-collar executives may peak in early morning hours when that audience is watching before commuting. The variables are granular and dynamic. "In performance, you don't have to go out and target lookalike audiences," he explains. "You run a campaign, you see what performs, and then you go back after the programs that delivered the highest results. By default, you're securing the inventory that's going to perform the highest for that specific product, and you can re-optimize every day if conditions change."
The manual infrastructure behind performance TV buying is staggering. Agencies negotiate individually with dozens of networks, each of which may have multiple sales teams covering different properties. Reallocating budget from one daypart to another requires phone calls, emails, and days of coordination. Daily optimization is theoretically possible, but practically impossible at scale. "It's very manual and very time consuming," Ellis says. "What we've developed is a system where you can put in one budget with campaign requirements and initial allocations across twenty networks, and the algorithms find the highest-performing available programming every day across all of them."
The strategic implication is significant. When the manual labor of reallocation is automated, buyers can step back and operate at a higher level, watching performance trends, tracking audience migration across networks and dayparts, and making decisions informed by heat maps that show which programming categories are heating up for specific product types. "AI is replacing the drudgery of the job and lets you be much more strategic. You can look at trends, you can chase audiences faster, and you can move money around more intelligently."
Ellis is direct about what this means competitively. The agencies that embrace performance optimization platforms will be able to demonstrate measurably superior results to prospective clients, creating a structural advantage over those still operating on the legacy manual model. "The agencies that engage in this are going to crush it," he asserts. "They can go to prospective clients and say, 'We can outperform the marketplace. No one else is doing this.' And it's free for agencies and advertisers. I never wanted to charge the buy side. I hate the tech tax."
He contrasts this with the connected TV ecosystem, where he estimates only about 65% of advertiser spend actually buys media after SSP, DSP, data, and ad-serving fees are deducted, with additional losses to bot traffic and fraud. Linear TV, for all its legacy inefficiencies, still delivers verified impressions at scale. With performance now representing nearly half of all linear revenue, up from roughly 10% two decades ago, Ellis believes the segment is too large and too strategic to keep running on a process that hasn't changed since the 1970s. "Programmatic is going to replace this legacy process. Those who are left behind, when you're talking about 50%-plus increases in performance and lower cost per outcomes, they'll catch up in a hurry. Because they're going to have to."
