Connected TV in 2026: Running CTV as a Performance Channel

Connected TV advertising crossed a structural line in 2026. US CTV ad spend is projected at roughly $37.95 billion this year, and — for the first time — CTV upfront commitments ($17.73B) are forecast to exceed primetime linear TV upfronts ($16.98B). The big screen in the living room is now a streaming, addressable, programmatic surface, and the dollars are following the eyeballs.

But scale is not the interesting part of the story. The interesting part is that most mid-market brands still treat CTV as a brand channel — a place to buy reach and hope — when the targeting, inventory, and measurement have quietly matured into something that can be run for performance. The brands that win in 2026 are the ones who stop measuring CTV with last-touch attribution it was never built for.

This guide is built for performance marketers and mid-market decision-makers. It maps the seven main ways to buy CTV by minimum spend and targeting data, sets realistic CPM expectations, frames the Amazon-versus-Trade-Desk fight that defines the buying landscape, and lays out the measurement shift — from attribution to incrementality — that separates brands who think CTV underperforms from brands who can prove it works.

The headline number is the easy part: eMarketer forecasts US CTV ad spend at roughly $37.95 billion in 2026. (Note that eMarketer's own models circulate slightly different figures — $37.70B appears in some articles — so treat any single number as directional, not audited.) The IAB's 2026 Ad Spend Forecast, built from a survey of 205 buy-side media-investment executives conducted between November 2025 and January 2026, projects CTV growth of +13.8% year over year — the second-fastest of any channel, behind only social media at +14.6%.

The structural milestone matters more than the topline. eMarketer projects that in 2026, US CTV upfront ad spending of $17.73 billion will exceed primetime linear TV upfront spending of $16.98 billion — a gap of about $750 million, and the first time streaming wins the upfront on the most-watched daypart. Combined, digital video plus CTV now hold roughly 23% of US ad spend, the largest share of any media channel, ahead of social (18.4%) and paid search (16.2%), while linear TV slides toward 11%.

Why now? The IAB attributes much of 2026's acceleration to a robust calendar of cyclical events — the Milan Cortina Winter Olympics, US midterm elections, and the FIFA World Cup — layered on top of a durable, ongoing shift in how Americans consume video. Reach has already moved: 89.5% of US households own at least one internet-connected TV device, and eMarketer projects 243.6 million US CTV viewers in 2026. Marketers reallocated an average of 36% of their linear TV budgets to CTV in 2025, and a Premion survey found 70% of advertisers plan to lift CTV investment by an average of 17% in 2026 — with the money coming primarily out of linear, display, search, and social.

Here is the single most actionable fact in CTV right now, and almost nobody frames it as an arbitrage opportunity. According to Nielsen, CTV accounts for 43.8% of total TV usage in the US. Yet as of 2025 it received only about 7.7% of total ad spend. Read those two denominators carefully — the first is share of TV viewing time, the second is share of all advertising dollars across every channel — but the directional message is unmistakable: where audiences spend their attention and where advertisers spend their money are badly out of sync.

That mismatch is a closing window. Every structural force in the market — addressable targeting, programmatic transaction (84% of CTV spend already trades programmatically, per Nielsen), and maturing measurement — is pulling spend toward viewing time. The advertisers who establish reach, creative, and measurement infrastructure while CPMs and competition are still relatively soft will own audience relationships that get more expensive to buy each quarter. This is not a "CTV is growing" observation; it is a timing argument. The gap rewards the early, not the eventual.

The longer arc reinforces it. Linear TV's share of global media spend has fallen from 41.3% in 2013 to 12.4% in 2026, and eMarketer projects linear TV ad spend will decline another 11% in 2026 across national and local. Money is not leaving television; it is moving within it, from a non-addressable surface to an addressable one. The question for a performance marketer is not whether to be on CTV — it is how to buy and measure it well enough to justify scaling spend ahead of the curve.

Most CTV explainers describe the buying landscape qualitatively and stop there. The table below is ours — a side-by-side decision matrix for mid-market advertisers, built from Adwave benchmarks, Vibe and improvado's Roku/DSP guidance, and platform documentation. Read it by constraint: start with the budget you can commit monthly, then with the targeting data that matters most to your category, and the right two or three rows will surface quickly. Minimums and CPMs are directional market figures, not contractual rate cards.

The practical read for most mid-market budgets: start self-serve. Roku's Ads Manager opens at roughly a $500 minimum campaign spend and is positioned for consumer brands in the $50M–$500M revenue band; Hulu's Ad Manager has a comparable entry point. These let you establish creative, audiences, and a measurement baseline before committing to a managed DSP. Move up to Amazon DSP or The Trade Desk — which typically expect $10,000–$25,000 per month — once you have data worth optimising against and a measurement framework that can prove the spend. Layering matters too: private marketplace deals account for roughly 47% of CTV spend versus only about 15% on the open exchange, so curated PMPs are where most premium inventory actually trades.

CTV CPMs are higher than social and digital video, and the premium buys something real: a non-skippable, full-screen, sound-on impression on the household's main television, completed almost every time. Across general programmatic inventory, CPMs commonly land in the $25–$65 range. Platform-specific benchmarks run roughly $25–$35 on Hulu, $20–$60 on Roku, and $10–$25 for YouTube CTV. Treat all of these as directional market estimates from bidding outcomes across many campaigns — not as official rate cards. Any individual buy will vary with daypart, audience, deal type, and inventory quality.

CPMs are also softening at the premium end, and that is a buyer's tailwind. After Amazon Prime Video launched its ad tier, competitive pressure pulled streaming CPMs down: eMarketer's modeling shows Netflix's average CPM falling from about $42.14 to roughly $31.05 by the end of 2024, and Prime Video's sliding from about $35.25 to roughly $28.01 over the same period. By Q2 2025, eMarketer forecast that only Netflix and Max would hold average CPMs above $30. These are vendor-sourced models (Digiday corroborates the downward direction), so read them as a trend, not as audited prices — but the trend is your friend as a buyer.

The CTV buying landscape is increasingly defined by a contest between two giants, and most coverage describes them as if they were interchangeable platforms. They are not. Amazon DSP and The Trade Desk are selling fundamentally different things, and which one fits depends on what you value more: authenticated first-party data, or neutral open optimisation.

The Trade Desk is also pressing its neutrality advantage at the infrastructure layer. On February 24, 2026 it introduced the Ventura Ecosystem, an open collaboration inviting TV OS makers and streaming platforms to build a more transparent CTV marketplace — with V (formerly VIDAA, powering 50M+ connected devices) and Nexxen as the first collaborators, and tools including OpenPath, Unified ID 2.0/EUID, OpenAds, and OpenPass. The strategic framing is explicit: independence from platform owners who optimise for their own inventory.

The decision is rarely either/or for a sophisticated advertiser. If your business is retail or DTC and shopper data drives your targeting, Amazon's authenticated reach is hard to replicate. If you value cross-platform optimisation and want to avoid handing your measurement to a walled garden that also sells inventory, The Trade Desk's open approach is the safer architecture. For retail brands weighing how much to concentrate in Amazon's ecosystem versus running independent media, our retail media buying strategy decision matrix works through the trade-offs in detail.

Here is the contrarian core of this guide. CTV does not underperform for most brands that say it underperforms — they are measuring it wrong. CTV is a non-click environment: a viewer watches a 30-second spot on the living-room TV, then later searches on their phone or buys on a laptop. Last-touch and multi-touch attribution, built on a chain of clicks, simply cannot see that path. They credit the final click and quietly zero out the CTV exposure that drove it. The channel that completes at 96% gets scored as if it did nothing.

The fix is the measurement shift already underway across sophisticated advertisers: from attribution to incrementality and media-mix modeling. Per eMarketer and TransUnion (July 2025), 52% of US brand and agency marketers already use incrementality testing, and 36.2% plan to invest more in it over the next year; separately, 46.9% plan to increase media-mix-modeling investment, and 27.6% already name MMM their single most reliable measurement methodology. These are the methods that read a non-click channel honestly — by measuring lift against a holdout, not by chasing a click that never happens on a television.

If you are standing up CTV measurement for the first time, sequence it deliberately. Use MMM to set the strategic budget split across channels, run incrementality tests to validate CTV's causal contribution on a cadence, and use clean rooms plus ACR to reduce household duplication and waste underneath both. For the mechanics of designing a clean lift test, our guide to incrementality testing methodology walks through holdout construction, and the MMM vs MTA decision framework explains exactly why a non-click channel like CTV breaks multi-touch attribution.

Performance marketers don't allocate budget to a channel in isolation — they allocate it against alternatives. The scorecard below puts CTV next to paid social (Meta), linear TV, and YouTube CTV across the dimensions that actually drive a budget decision. It is a strategic comparison, not a rate card: the goal is to make the trade-offs explicit so you can see where CTV earns a place in the mix and where another channel does the job better.

The pattern that emerges is clear: CTV's edge is household-level targeting on a guaranteed-completion, big-screen impression, paid for with incrementality rather than clicks. That is a different job than paid social's user-level, click-attributable, lower-CPM performance engine. The strongest 2026 media plans don't pick one — they use CTV for addressable reach and brand lift that incrementality can prove, social and search for lower-funnel capture, and YouTube CTV as a reach extension on the largest single CTV audience: 171 million people watch YouTube on TV screens monthly in the US.

Pulling it together, here is the operating sequence we use with mid-market advertisers entering or scaling CTV. It is deliberately ordered: get the measurement scaffolding right before you scale spend, not after.

Two emerging plays deserve a place on the roadmap. First, interactive and shoppable CTV formats: eMarketer data cited by StackAdapt reports these increase viewer engagement time by 71 seconds and lift brand recall by 36%, and 41.8% of US marketers already use them — they turn a passive impression into a measurable interaction. Second, retail media on CTV, which reached $4.46 billion in 2025 and is projected to more than double to $9.03 billion by 2029, fusing shopper data with big-screen reach. Both are where CTV's performance story gets sharper, and both sit naturally inside an integrated paid media program rather than as a standalone TV buy.

The forward view: as measurement maturity catches up to viewing time, the 7.7%-of-ad-spend figure should climb toward CTV's share of attention, and CPMs at the premium end will keep compressing under inventory competition. The advertisers who will benefit most are the ones who built the measurement muscle early — who can already prove CTV lift with a holdout — because they will scale into a maturing channel with conviction while competitors are still arguing about last-touch numbers. That early infrastructure, not the media buy itself, is the durable advantage.