Connected TV (CTV) Advertising in 2026: How It Works, What It Costs, and Where to Buy It

Connected TV advertising places video ads inside streaming content watched on an internet-connected television. It combines the full-screen attention of TV with the audience targeting, frequency control, and conversion tracking of digital media.

That much is settled. The imbalance shows up afterward, when the buyer negotiates a rate knowing less than the seller does and then grades the outcome using tools the seller built.

Correcting for that is the job of the rest of this page, starting with the two numbers sellers are least forthcoming about, what a thousand impressions should cost and how much budget it takes before the result means anything.

TL;DR: Key takeaways

Six things to know before reading further.

• The market is worth somewhere between $29 billion and $38 billion this year, depending on whose definition you use. Both figures come from credible sources counting different things, and the difference changes how you read every forecast in the channel. We explain why below.
• Inventory tier drives price more than platform choice does. Free ad-supported services and premium direct buys span a fourfold range, and most of what sets your rate is decided before the buy goes live. Full CPM ranges below.
• CTV's fraud exposure now matches the open web. Invalid traffic on US CTV has climbed sharply over the past year, and the channel's reputation for being materially safer than display is out of date.
• The channel suits advertisers who can fund a readable test. Platform minimums are lower than the budget you actually need to tell a working audience from a lucky fortnight.
• Streaming passed a milestone in early 2026, taking 46.6% of all ad-supported television viewing in the US.
• Getting started needs three things: a buying route, a conversion tracking setup that survives the gap between TV exposure and purchase, and creative built for a screen across the room.

What is CTV advertising?

CTV advertising delivers targeted video ads to viewers through internet-connected television devices while they stream content. Traditional broadcast commercials play to whoever happens to be watching. Connected TV ads can be customized based on household data, viewing patterns, and cross-device behavior.

Two neighbors watching the same show at the same time might see completely different commercials, one for a luxury car and the other for a local pizza shop, based on their respective profiles and interests.

The process works through dynamic ad insertion, where streaming platforms and apps serve different ads to different households in real time. Advertisers buy this inventory either programmatically through automated bidding systems or directly from publishers like Hulu or Peacock.

Connected TV refers to any television set that connects to the internet and streams digital video content. Smart TVs with built-in connectivity qualify, as do regular televisions connected through external streaming devices or gaming consoles. The internet connection is what allows data collection and targeted ad delivery.

Devices that support connected TV

The connected TV ecosystem runs on four types of hardware, and campaigns typically reach households across all of them.

• Smart TVs from manufacturers like Samsung, LG, and Vizio build internet connectivity directly into their sets, requiring no additional hardware.
• Streaming devices including Roku, Amazon Fire TV Stick, Apple TV, and Google Chromecast bring CTV capabilities to any television with an HDMI port.
• Gaming consoles serve double duty as entertainment hubs, with PlayStation and Xbox users spending significant time streaming video.
• Modern cable and satellite boxes from providers like Comcast now include streaming apps, blurring the line between traditional and connected TV.

CTV vs. OTT vs. linear TV: what's the difference?

These three terms describe overlapping things, which is where most of the confusion starts. 

• Linear TV is scheduled broadcast delivered by cable, satellite, or antenna. 
• Over-the-top (OTT) is any video delivered over the internet to any device. 
• Connected TV sits inside OTT and covers internet-delivered video watched on a television screen.

All CTV is OTT. Not all OTT is CTV. Watching Disney+ on a phone is OTT alone; watching the same show on a smart TV is both. The distinction carries real weight for buyers, because the television environment produces different viewer behavior: more attention, less skipping, and frequent co-viewing.

💡 For a fuller breakdown of how the two categories overlap, see our guide to OTT vs CTV.

How big is the CTV market, and why the numbers disagree

Search for the size of the US connected TV ad market and you will find figures between roughly $29 billion and $38 billion for the same year. Both ends come from credible forecasters. They disagree because they are counting different things, and neither publishes a reconciliation against the other.

Neither of these figures is wrong, and the practical lesson is to treat any single number as directional and to ask what a forecast counts before comparing it with another one. A CTV market size quoted without its definition attached is telling you less than it appears to.

The IAB split contains the more useful story for planners. Total US digital video spend reaches $81.9 billion in 2026, exceeding 60% of combined TV and video ad spend for the first time. Social video overtook CTV in absolute dollars back in 2025, and in 2026 it grows faster too, 13% against CTV's 11%, which IAB flags as a first. Buyers treating CTV as the automatic home for video budget should check that against where the incremental dollars are actually going.

One structural milestone is worth noting alongside the topline. eMarketer projects US CTV upfront commitments of $17.73 billion in 2026 against $16.98 billion for primetime linear, which would be the first year streaming wins the upfront on the most-watched daypart.

The viewing data underneath the forecasts is less contested. Nielsen's Q2 2026 Ad Supported Gauge put streaming at 48.2% of ad-supported television viewing, up 1.6 share points on the quarter, with broadcast at 26.6% and cable at 25.2%. Ad-supported platforms carried 71.5% of total viewing. On the wider monthly measure, streaming reached 48.5% of all US television viewing in June 2026, against 47.5% in December 2025.

Two caveats travel with those numbers. 

• Nielsen has not yet migrated the Ad Supported Gauge to the ARF DASH-based universe estimates, a change planned for the fall that will move the figures. 
• And the Gauge counts viewing through a television screen only, so streaming on phones and laptops sits outside it entirely.

How does CTV advertising work?

CTV advertising runs on an auction that resolves in the time it takes a stream to load. When someone starts streaming, the app sends a request containing anonymized household information to ad servers, which auction the opportunity to show an ad. The winning advertiser's video plays during the break, and performance data flows back for analysis and optimization.

This happens thousands of times per second across millions of households. The technology stack involves demand-side platforms (DSPs) where advertisers manage campaigns, supply-side platforms (SSPs) that publishers use to sell inventory, and identity and data platforms that organize audience information.

Programmatic buying in CTV

Programmatic buying automates the purchase and placement of CTV ads through real-time bidding. Rather than negotiating rates manually with each publisher, advertisers use a DSP to set parameters: target audience, budget, frequency caps, and performance goals. The platform then bids on available inventory across multiple streaming services at once.

The sequence resolves in the moment before the stream resumes. A viewer presses play, an ad request hits the marketplace, algorithms evaluate whether the viewer matches the target criteria, and the highest bidder's ad plays.

Programmatic is now the default rather than the alternative. eMarketer put programmatic at 84.2% of US CTV video ad spend in its 2025 forecast. Direct publisher deals still exist and still count for premium inventory, but they are increasingly executed as programmatic guaranteed rather than as insertion orders.

Deal types determine how much control you trade for how much access. 

• Open exchange buying gives the widest reach at the lowest cost and the least transparency. 
• Private marketplaces restrict the auction to invited buyers on defined inventory.
• Programmatic guaranteed fixes both the price and the volume in advance, which is how most premium streaming inventory is now sold.

💡 Our guide to CTV media buying covers the mechanics of each in depth.

CTV ad formats

Connected TV inventory comes in a handful of formats, and the choice affects both price and what you can measure.

• Pre-roll ads play before content begins, typically running 15 or 30 seconds.
• Mid-roll ads interrupt content at natural breaks, mimicking traditional commercial pods but with lighter ad loads, generally two to four minutes against roughly eight minutes on linear TV.
• Interactive ads add a response mechanism to the spot. QR codes let viewers connect their phones, while shoppable overlays display product information without interrupting the video. Innovid has measured interactive CTV formats generating more than a 600% lift in engagement against standard pre-roll.
• Pause ads and home-screen placements sit outside the content stream entirely, appearing when a viewer pauses or browses. Roku, Hulu, and Amazon all sell them, and they price differently from in-stream inventory.

Creative length is a pricing lever as well as a creative one. Fifteen-second spots generally clear at lower CPMs than thirty-second spots on the same inventory, though the difference varies by publisher and by whether the buy is programmatic or direct.

Targeting and personalization

Targeting precision is the main reason budget moves from linear TV to CTV, and it operates on five broad approaches that campaigns usually layer rather than choose between.

First-party data integration is where CTV separates from broadcast most sharply. Brands upload customer lists to build lookalike audiences or retarget existing customers, bringing CRM-level targeting to the television screen.

Contextual targeting matches ads to content themes without using personal data. Amazon has reported engagement rates up to 25% higher for contextually targeted placements than for demographic targeting alone, and the approach has gained ground as identity signals have become harder to rely on.

A note on what targeting costs. Every layer you add narrows the addressable pool and raises the price of reaching it, which is the mechanism behind the CPM ranges in the next section. Precision is a purchase, not a feature.

Benefits of connected TV advertising

The case for CTV rests on four advantages that linear television cannot offer, and one of them is worth quantifying before the rest.

Less waste

Linear television buys reach large audiences with limited control over who is inside them. One analysis of a diaper brand's linear spend estimated that approximately $60 million of a $75 million campaign reached households that do not buy diapers.

Household-level targeting closes most of that gap. It does not close all of it, since household data has its own accuracy limits, but the difference between broad demographic buying and addressable buying is the largest single source of efficiency in the channel.

Attention on the biggest screen

The full-screen, sound-on environment of television commands a kind of attention that smaller screens rarely match. In a tvScientific consumer survey, 46% named TV and streaming as the most memorable ad environment, against 33% for social and 9% for websites.

Co-viewing adds to this. A single impression on a living-room screen frequently reaches more than one person, which is an advantage that CTV measurement is still learning to count properly. Reported impressions therefore understate real reach on family and sports content, while overstating it in households where the television runs unwatched. Neither correction is applied by default in most reporting, so campaigns that depend on precise reach numbers should ask how co-viewing is modelled before comparing CTV reach against any other channel.

Cross-device measurement

Identity resolution links the living-room television to the smartphones, tablets, and laptops in the same household, which enables closed-loop measurement that traditional television never supported.

A viewer sees a CTV ad for vacation deals on Sunday night. By Wednesday they have booked a trip, moving from TV to work computer to tablet to phone along the way. Linear TV would miss that journey entirely.

Optimization in flight

Programmatic delivery brings mid-campaign adjustment to television buying. Bidding algorithms move budget toward better-performing placements daily rather than at the end of a flight. Dynamic creative optimization tests different messages and calls to action to find the combinations that work. Budget pacing distributes impressions evenly instead of leaving a scramble at the end of the month.

How much does CTV advertising cost in 2026?

Almost nobody publishes CTV pricing, which makes the question harder to answer than it should be. The figures below come from 2026 benchmark data and from publicly disclosed platform rates, with the caveat that CTV pricing is negotiated and varies by flight, season, and buyer scale.

💡 CTV is bought on a CPM basis, the cost per thousand impressions. Our explainer on what CPM means in TV advertising covers the mechanics.

CTV has split into two pricing tiers. FAST and standard ad-supported inventory runs roughly $15 to $40, while premium direct buys run roughly $35 to $65, according to 2026 benchmark analysis built on Solomon Partners media data. Where a given campaign lands inside those bands depends far more on targeting and deal type than on which service the ad runs against.

Platform-level rates give a more practical picture.

For reference, broadcast primetime cleared around $43.50 CPM in the 2024-25 upfront, with 2026 benchmarks putting premium primetime at $40 to $60. Premium CTV therefore prices at rough parity with the linear inventory it competes against, while standard CTV undercuts it.

Two further anchors are worth knowing. Paramount publishes a $25 to $65 industry benchmark on its own site while offering entry pricing from $7 CPM through its self-serve platform, a spread that shows how far buying route alone moves the number, and one published by a party with an interest in the comparison. Competitive supply has also pushed premium pricing down rather than up: eMarketer's modeling shows Netflix's average CPM falling from roughly $42 to around $31 across 2024, with Prime Video sliding from about $35 to $28 over the same period as Amazon switched ads on by default. These are vendor models rather than audited rates, so read them as direction rather than as a rate card.

What moves your CPM

Two campaigns buying the same inventory can pay very different prices. Six factors account for most of the variance.

• Targeting depth. Broad demographic targeting sits at the bottom of the range. Each layer you add, from first-party audience matches to behavioral segments to tight geography, narrows the addressable pool and lifts the clearing price.
• Geography. ZIP-code and DMA-level targeting costs more per impression than national delivery, and dense metro areas cost more than rural ones.
• Deal type. Open exchange is cheapest, private marketplace sits in the middle, programmatic guaranteed costs most and delivers most reliably.
• Content and daypart. Live sports, primetime, and premium originals price above library content and off-peak inventory.
• Frequency caps. Tight caps reduce the pool of impressions you can win, which raises the clearing price of the ones you do.
• Seasonality. Q4 and major event windows compress supply. See the 2026 event calendar for the three that will move pricing this year.

What you need to spend

Budget floors are set by measurement rather than by platform minimums. Self-serve DSPs set an account minimum, listed in the platform comparison, but the more useful threshold is the one at which results become readable.

At $25,000 to $50,000 a month, a campaign generates enough impressions and conversions to separate real performance differences from noise inside a reasonable test window. Below that, you will get delivery and reporting, but not a result you can act on. Advertisers with smaller budgets are usually better served by concentrating spend on a narrow geography or a single publisher rather than spreading a thin buy across the whole ecosystem.

CTV advertising platforms compared

Where you buy determines what inventory you can reach, what data you get back, and how much of your budget reaches the publisher. Four routes cover almost all CTV buying.

• Self-serve DSPs suit teams with programmatic experience who want control over bidding, supply paths, and measurement. They give the widest inventory access and the most data, at the cost of requiring someone who knows how to use them.
• Managed service providers handle execution and bring CTV expertise for a percentage over media. The trade is expertise for margin, and the quality of the outcome depends heavily on whether the provider's incentives point at your KPIs or at their own supply.
• Direct publisher buys guarantee placement and brand safety on named inventory. They cost more and confine reach to one platform, which suits advertisers who need a specific premium audience rather than scale.
• Publisher self-serve tools have opened CTV to local and small advertisers at genuinely low minimums. Reach is limited to that publisher's footprint, and targeting and measurement are correspondingly basic.

Where the audiences are

Platform reach figures come mostly from the platforms themselves and should be read as directional. Netflix reported its ad tier passing 250 million monthly active viewers globally at its May 2026 upfront, up from 190 million six months earlier. Amazon claimed 315 million monthly ad-supported viewers worldwide at its November 2025 unBoxed event, including more than 130 million in the US across 16 countries, a figure that reflects Prime Video's decision to make ads the default rather than an opt-in tier. Both are internal counts on methodologies the companies set themselves, and neither is verified by a third-party measurement firm.

For a figure not supplied by the platforms themselves, the Video Advertising Bureau's 2026 streaming report puts the US ad-supported streaming audience at 209.4 million, up 27% on 2023. That number originates with eMarketer rather than with VAB's own measurement, so it is independent of the platforms but not of the forecaster quoted earlier in this guide.

The practical consequence of this scale is duplication. Buying Netflix, Prime Video, Hulu, and a FAST bundle in the same flight will reach many of the same households repeatedly unless frequency is managed across the whole plan rather than within each platform.

Is CTV advertising worth it?

The honest answer depends on arithmetic that most guides skip. CTV costs several times more per impression than display or social. It earns that premium back through completion, attention, and targeting precision, or it does not.

Work it as a break-even. At a $30 CPM, $30,000 buys roughly one million impressions. If your product carries a $200 margin, the campaign needs 150 conversions to break even, which is a 0.015% conversion rate on impressions served. For a considered purchase with a long path to sale, that is achievable. For a low-margin, low-consideration product, it usually is not, and the money works harder in a channel with cheaper reach.

Two pieces of evidence support the upside. Analytic Partners' ROI Genome data, drawn from more than 750 brands across 45 countries, puts CTV's ROI 30% above other marketing channels while CTV takes an average of just 7% of total spend. On that basis the firm recommends 10% to 20% as a floor. In one Trade Desk case study, an extended warranty brand pairing CTV with Walmart Connect shopper data doubled revenue year over year and lifted ROAS 67%, though that comparison runs against a prior campaign bought through a different route, so the buying method accounts for part of the gain.

Scale is no longer the barrier it was. Premion reports a 70-store convenience chain using radius-targeted CTV for two grand openings, lifting sales on promoted items by 930%, 77% and 69%. A performing arts theatre sold out its season subscriptions and drew 502 website visits inside a week. Both were geographically tight buys on modest budgets, which is the shape that works for local advertisers, though the opening and launch moments were doing some of the work alongside the media.

The channel is worth it when three things are true: 

1. your margin supports premium CPMs, 
2. you have creative that works with sound on and a screen at distance, and 
3. you can measure something beyond last-click. Miss any of those and the answer changes.

When CTV is the wrong buy

Four situations come up often enough to name, and in each of them the budget usually performs better somewhere else.

• Your budget cannot fund a readable test. Covered above, and it is the most common reason CTV campaigns disappoint. A $5,000 monthly buy spread across four platforms produces delivery reports and no learning.
• Your only measurement is last-click. CTV rarely converts on the device that served the ad. Attribution models that credit only the final touch will show CTV contributing almost nothing, and the campaign will be cut before it has been fairly assessed. Without view-through windows or incrementality testing in place before launch, the channel cannot prove itself.
• Your creative is a repurposed social cut. Vertical video with burned-in captions, designed for a muted phone held at arm's length, performs badly on a horizontal screen watched from across a room. Production cost has to be part of the CTV budget rather than an afterthought.
• You need immediate response at low cost. CTV builds consideration and assists conversion. Advertisers who need a same-week direct response at a low cost per acquisition are generally better served by search and shopping formats, with CTV added once the lower funnel is saturated.

One further caution applies to how CTV performance gets reported. DoubleVerify has found that over a third of CTV impressions are delivered in environments capable of playing ads while the television is powered off, a problem it estimates wastes around $1 billion a year. The ad still records as served and completed. Any assessment of whether CTV is working should account for that gap, and viewability verification should be specified at the buy rather than assumed.

CTV metrics and measurement

Measuring CTV performance requires rethinking traditional television metrics while taking advantage of digital attribution. Five measures carry most of the weight.

• Video completion rate is the channel's foundational metric. Peer39 gives the industry standard for non-skippable CTV as 90% to 97%, and recorded 97.51% on its own quality-controlled inventory in Q2 2026. Read any completion figure alongside the viewability caveat below, since a completed impression and a watched impression are not the same thing.
• Reach and frequency gain real precision. Instead of panel-based estimates, advertisers see how many unique households were served, though deduplicating across platforms remains difficult.
• Attribution and conversion tracking follow viewers from exposure through site visits and purchases, using view-through windows of seven to fourteen days rather than click-based models.
• Brand lift studies measure changes in perception and purchase intent through survey research. A GumGum campaign for Compare the Market lifted brand favourability eight points and likelihood to use seven points against a control group, with top-of-mind awareness up six points among higher-value audiences.
• Incrementality testing isolates CTV's true contribution by holding out geographies or audience segments and comparing outcomes. It is the most reliable method available and the least used.

Cross-device attribution

Identity resolution is what makes CTV measurable at all, since almost no one converts on the television. Two techniques extend tracking beyond digital actions.

• Store visit attribution uses mobile location data to connect exposure with physical retail trips, which is how CTV gets credited for offline sales.
• Multi-touch attribution distributes credit across the full purchase path rather than assigning it to the last click, recognizing the awareness and consideration that CTV builds before another channel closes.

Both depend on matching a household to its devices, and match rates vary considerably by provider. Ask for the match rate before accepting the attribution report built on top of it.

Challenges in CTV advertising

CTV's advantages come with three structural problems that buyers have to plan around rather than solve.

Fragmentation of platforms and measurement

Hundreds of streaming services, apps, and content providers operate across the CTV ecosystem, each with its own advertising policies, creative specifications, targeting capabilities, and reporting standards. Comparing performance across them means reconciling different impression definitions, attribution windows, and reach methodologies.

The walled garden problem compounds this. Major platforms guard their user data closely, which prevents the unified audience view that buyers expect from digital channels.

Frequency capping and ad repetition

Ad repetition is the most common viewer complaint about CTV, and it follows directly from fragmentation. The same household can receive the same ad across multiple services, well past the point of diminishing return, because no service can see the others' delivery. Industry practice puts optimal exposure at roughly three to seven impressions per week, a working convention rather than a published finding; unmanaged multi-platform buys routinely exceed it.

Identity resolution tools and single-platform concentration both reduce the problem. Universal frequency capping across the open CTV ecosystem remains unsolved.

Ad fraud in the CTV ecosystem

The channel's reputation for safety is out of date, and the gap between that reputation and current conditions is the most important correction in this guide.

Pixalate's Q2 2026 benchmarks put invalid traffic on US CTV at 25%, level with desktop and mobile web and below mobile apps at 39%. Global CTV invalid traffic runs slightly higher at 26%. The trend is what should concern buyers: the same US measure sat at 18% in Q2 2025 and 19% in Q4 2025 before reaching 25% by mid-2026. Invalid traffic is a broader category than fraud, covering bots and non-human traffic as well as deliberate deception. The channel was routinely described as materially safer than the open web. That is no longer accurate.

DoubleVerify's 2026 Must-CTV report explains the change. Fraud schemes and their variants targeting CTV rose 140% globally in Q1 2026 against the same quarter a year earlier, following a tenfold increase in fraudulent CTV apps detected during 2025 and more than 50 distinct bot attacks identified that year. The firm estimates roughly $1.8 million lost per billion unprotected impressions.

The same analysis contains the useful part. Fraud rates in protected campaigns stayed below 1%, against nearly 9% in unprotected ones. The exposure is almost entirely a function of whether verification and supply-path controls are in place before the campaign launches, rather than of the channel itself.

Two techniques account for most CTV fraud. 

• Device spoofing falsifies the device identifiers in a bid request so that cheap or nonexistent inventory is sold at CTV prices. 
• Server-side ad insertion (SSAI) spoofing exploits the server-based delivery model to fabricate impressions that no viewer ever saw. 

Both are hard to detect from campaign reporting alone, which is why filtering has to happen at the supply level. 

💡 Our guide to CTV ad fraud covers detection in more detail.

Brand safety and brand suitability are separate problems, and CTV handles them unevenly. 

• Safety is largely solved: major streaming services vet their programming, and the walled-garden structure of premium apps keeps genuinely harmful content out. 
• Suitability is harder, because content-level signals in CTV bid requests are often thin or missing entirely. A buyer can know which app served the impression without knowing which programme ran inside it, which makes adjacency control weaker than it is on the open web. Advertisers in sensitive categories should ask what content metadata their supply actually carries before assuming that a premium app guarantees a suitable placement.

Supply-side filtering is the practical defense. Smart Supply applies AI-powered filtering that removes fraud, invalid traffic, and inefficient placements before inventory reaches the buyer, with every deal ID built around campaign KPIs and no DSP or SSP bias in the sourcing. Combined with pre-bid verification, it addresses the exposure at the point where it originates rather than reporting on it afterward.

The 2026 event calendar

Three cyclical events land in 2026 and all three will compress CTV inventory and raise clearing prices in their windows.

IAB's 2026 Outlook Study projects US CTV ad spend growing 13.8% this year while linear TV declines 1.7%, against total US ad spend growth of 9.5%. Strip out the cyclical events and that total growth figure falls to between 7.1% and 7.8%, which is the clearest available measure of what the midterms, the Winter Olympics, and the FIFA World Cup are adding to the market.

• The US midterm elections absorb enormous volumes of local CTV inventory in contested states from roughly September onward. Political buyers pay above market and buy geographically, which means advertisers targeting the same DMAs should expect both higher CPMs and reduced availability. Plan flights around it or budget for the premium.
• The Winter Olympics concentrate a large, ad-supported streaming audience into a short window with high co-viewing. Inventory is expensive and mostly sold in advance through the rights holder.
• The FIFA World Cup, hosted across North America, delivers the largest live streaming audiences of the year. Live sports inventory prices at the top of the CTV range and sells out early.

For advertisers not competing for those audiences, the same dynamic creates opportunity. Non-event inventory in the surrounding weeks frequently clears below normal as attention and budget concentrate elsewhere.

How to get started with CTV advertising

Launching a first CTV campaign comes down to four decisions, taken in order.

1. Choose a buying route

Match the route to your team rather than to your budget. In-house programmatic experience points to a self-serve DSP; its absence points to a managed partner or a publisher self-serve tool. The comparison above covers the trade-offs. Whichever route you take, ask what the platform's supply path looks like and how many intermediaries sit between your budget and the publisher, because that gap is where efficiency disappears.

2. Set up measurement before launch

Pixels, conversion tracking, and a control group have to be in place before the first impression serves. Retrofitting measurement onto a running campaign produces numbers nobody trusts. Define the view-through window in advance, agree which conversions count, and hold out a geography or audience segment so that incrementality can be assessed at the end.

Advertisers who need this consolidated across platforms can use Elevate to centralize metrics across integrated DSPs and map cross-screen paths to conversion, with optimization set against revenue and actual conversions rather than platform-reported metrics.

3. Allocate and pace the budget

Spread inventory rather than concentrating it, and reserve room to learn. A workable starting split puts the majority of spend on major platforms with proven reach, a meaningful share on broadcaster apps, and a smaller allocation on emerging inventory being tested for efficiency. Hold back 10% to 15% for testing creatives, audiences, and platforms, a planning convention rather than a benchmarked figure.

Pacing choices depend on the goal. Steady distribution suits always-on brand building; flighting concentrates weight around launches or promotions; dayparting into peak viewing hours, roughly 7pm to 11pm, maximizes reach per dollar at the cost of higher competition.

4. Build creative for the room

The screen is large, the viewer is several feet away, and the sound is on. Use cinematic visuals, minimal text at large sizes, and logos legible from across the room. Invest in the audio, since most CTV viewing happens with sound on and the voiceover carries more of the message than it would on mobile. Where the format allows, vary creative by daypart and context.

Specifications vary by publisher, but a few requirements are close to universal and worth building to from the start. 

• Deliver in 1080p at minimum, with 4K masters held in reserve for premium placements. 
• Keep to standard durations of 15 and 30 seconds, since non-standard lengths are rejected by many streaming publishers outright. 
• Audio should be mixed to broadcast loudness standards rather than to the louder levels common in social video, because streaming services normalize and a hot mix simply gets turned down. 
• Allow lead time for approval: premium publishers commonly require creative one to two weeks before launch, and a rejection late in that window can cost the whole flight.

CTV trends in 2026

Five developments are worth tracking this year. Our CTV advertising trends guide covers each in depth.

• Converged TV planning replaces channel-by-channel buying. Linear and streaming are increasingly planned as a single video buy with deduplication and incrementality built in from the start, rather than as separate line items.
• Retail media meets CTV. Retailers are bringing first-party purchase data into CTV targeting, letting buyers target actual purchase behavior and close the loop back to in-store and online sales.
• Measurement standards begin to stabilize. Wider adoption of the IAB's CTV Conversion API and clean-room infrastructure is narrowing the gap between platform-reported results and independently measured outcomes.
• Interactive and shoppable formats move from pilot to line item. QR codes, dynamic product overlays, and one-click purchase paths are shortening the distance between the ad and the transaction.
• AI-driven contextual targeting matures. Content-level analysis of dialogue, visuals, and themes is replacing demographic proxies in inventory where identity signals are weak. For more on this, see the rise of AI in TV advertising.

Best practices for CTV campaigns

Four habits separate campaigns that read well from campaigns that merely deliver.

• Define KPIs before launch. Set primary and secondary metrics tied to business objectives, and account for CTV's assist effect by tracking influenced conversions rather than direct ones alone.
• Layer first-party and contextual data. Combining owned customer data with content-level matching produces tighter audiences without depending on identity signals that are becoming less reliable.
• Plan CTV inside the full funnel. The channel builds awareness and consideration that other channels convert. Cross-device retargeting off CTV exposure is where much of the measured return comes from.
• Cap frequency across the plan, not within each platform. Given the duplication problem described above, plan-level frequency management is the single highest-return operational discipline in CTV buying.

Conclusion

Connected TV has moved from an experimental line item to a core video channel, and the questions buyers ask about it have moved with it. Pricing, platform choice, and measurement now decide outcomes more than the decision to use the channel at all.

Three things separate campaigns that work from campaigns that spend. 

1. Fund the test properly, because CTV punishes budgets too thin to read. 
2. Verify the supply, because the channel's fraud exposure now matches the open web and the difference between protected and unprotected campaigns is the difference between a working buy and a wasted one. 
3. Measure beyond the last click, because CTV's contribution happens upstream of the conversion and last-touch models will not find it.

The advertisers doing well in CTV in 2026 are not the ones with the largest budgets. They are the ones who decided in advance what they were buying, what it should cost, and how they would know if it worked.

Talk to AI Digital about building a CTV plan with the supply quality and measurement to back it.