Connected TV Ads for DTC Brands: A 2026 Performance Playbook

Quick answer: Connected TV (CTV) ads are video ads served through internet-delivered streaming apps like Netflix, Amazon Prime Video, Roku, Hulu, Disney+, and YouTube CTV on a television screen. In 2026 US CTV ad spend hit USD 37.95B, up 18.4 percent year on year, with 84 percent bought programmatically. DTC brands from USD 5M to USD 200M ARR now allocate 5 to 15 percent of paid budget to CTV, mostly to lift branded search and new customer share, not to hit last-click ROAS.

Every DTC founder we work with in 2026 is asking the same two questions: is CTV worth it below USD 20M ARR, and how do we measure it without a click. Across 3 DTC apparel accounts running USD 8K to USD 18K/mo in CTV between March and June 2026, blended CAC dropped 12 to 19 percent within 60 days once we added a geo holdout.

By Nand Patel, Founder, New Public Media.

What Are Connected TV Ads and Why DTC Brands Are Shifting Spend in 2026

Connected TV ads are video ads delivered through a smart TV, streaming stick, or gaming console via an internet-connected app, as opposed to linear cable or broadcast. OTT (over the top) is the delivery method, CTV is the device. So a Hulu ad on a Roku stick is CTV, and the same Hulu ad on a phone is OTT mobile.

The 2026 spend picture: US CTV ad revenue hit USD 37.95B (+18.4 percent YoY) per Insider Intelligence 2026 CTV forecast, 84 percent of that inventory transacted programmatically, and CTV upfronts passed primetime linear commitments for the first time in the 2026/27 season. Completion rates on premium CTV inventory sit at 95 to 98 percent, versus 30 to 45 percent for YouTube in-stream skippable.

DTC brands from USD 5M to USD 200M ARR are moving 5 to 15 percent of paid budget into CTV because Meta and Google reach ceilings are compressing (see why paid social keeps getting more expensive) and CTV is one of the last channels where a new impression is genuinely incremental.

Key takeaway: In 2026 CTV is not a brand tax, it is a performance channel that lifts branded search, direct, and new customer share when you measure it with incrementality instead of last-click.

Connected TV Ads vs Meta and Google: Where CTV Wins and Loses

CTV wins on incrementality, new customer share, and completion rate. It loses on last-click ROAS, targeting granularity, and speed of iteration. If your CFO measures every channel by same-session ROAS, CTV will lose every quarter, even when it is the reason branded search is growing.

The right frame is a flywheel: Meta and creator content drives cheap discovery, Google captures the resulting search demand, and CTV adds a reach layer that lifts both without cannibalising them.

Key takeaway: CTV is priced 2 to 4x above Meta on CPM but pays back through branded search lift and new customer share, not through platform ROAS reports.

The Six CTV Platforms Worth Testing in 2026 (Netflix, Amazon Prime Video, Disney+, Roku, Hulu, YouTube CTV)

Not every CTV platform is worth a test. The six that matter for DTC in 2026 are Netflix, Amazon Prime Video, Disney+, Roku, Hulu, and YouTube CTV. Everything else (Peacock, Max, Paramount+, Tubi, Pluto, Fire TV FAST) is reach extension inside a DSP, not a standalone buy.

Netflix

Netflix ended Q1 2026 with 94M ad-tier subscribers and 190M MAU on the ad tier, per Netflix Q1 2026 letter to shareholders. Advertiser count grew 70 percent YoY to 4,000+ brands, and Netflix is targeting USD 3B in 2026 ad revenue. Since Q1 2026, programmatic access is open via DV360, The Trade Desk, Amazon DSP (Q2 2026), and Yahoo DSP. Direct IO minimum is USD 45K, programmatic self-serve minimum is USD 500/day.

Amazon Prime Video

Amazon Prime Video removed the self-service minimum spend at Amazon Ads unBoxed 2025 in October 2025. Since Q1 2026, any DTC brand with an Amazon DSP seat can launch Prime Video for as little as USD 500 test spend. Prime Video sits inside Amazon DSP with retail purchase signals attached, which is why we usually run it alongside our Amazon DSP program (see Amazon DSP for DTC brands).

Disney+, Hulu, Roku, YouTube CTV

Disney+ ad tier: 190M global users, programmatic via The Trade Desk and DV360, USD 25 to 45 CPM. Hulu: premium AVOD, USD 25 to 55 CPM. Roku: 90M+ active accounts, cheapest premium CTV inventory at USD 20 to 40 CPM, plus Roku FAST at USD 15 to 25. YouTube CTV: 250M+ monthly viewers on TV screens, bought through Google Ads Video reach campaigns, USD 12 to 22 CPM.

Key takeaway: Start with Amazon Prime Video via DSP (lowest floor, retail data) plus Roku FAST for cheap reach, add Netflix programmatic once you can prove incremental lift.

CTV Ads CPM and CPA Benchmarks Across Platforms and Verticals

CTV CPMs sit between USD 15 (broad AVOD/FAST) and USD 65 (Netflix direct, premium AVOD direct IO). On 11 DTC accounts we ran between January and June 2026, blended CTV ROAS landed between 2.0x and 3.5x once incrementality was factored in, versus 1.1x to 1.6x on platform last-click.

Response benchmarks on strong DTC creative: 0.2 to 0.6 percent site-visit conversion, 0.02 to 0.08 percent same-session purchase, and a 6 to 14 percent lift in branded search within 14 days at USD 8K+ monthly CTV spend.

Key takeaway: Blended CTV ROAS of 2.0x to 3.5x is realistic in 2026, but only if you refuse to judge the channel on platform ROAS reports.

How to Set Up Your First CTV Campaign in 30 Minutes

You do not need an agency to launch a CTV test, you need a DSP account, one 30-second cut, and a geo holdout. Here is the 8-step launch:

1. Pick a DSP. Amazon DSP, The Trade Desk, DV360, or Yahoo DSP for USD 5K+ monthly tests. Vibe, MNTN, or Tatari for USD 1K to 5K self-serve tests without a trading desk.
2. Import audiences. Push your Meta CAPI purchasers, Klaviyo 90-day buyers, and site visitors into the DSP via LiveRamp or direct integration.
3. Build a lookalike seed of 90-day purchasers and cap age of source to 30 days for freshness.
4. Frequency cap at 3 to 5 impressions per household per week, and 1 per household per day. CTV frequency waste is the number one budget killer.
5. Upload 15s and 30s creative (never 6s or 60s on CTV, wrong slot lengths).
6. Enable ACR (automatic content recognition) exclusions where the DSP supports it (TTD, Amazon DSP) so you do not pay to reach the same household that just watched your ad on another app.
7. Set an initial spend floor of USD 100 to USD 300 per day for a 14-day read, split 60 percent premium AVOD, 25 percent Netflix or Prime Video, 15 percent FAST.
8. Create a matched pair of geo holdout DMAs (or postcodes in UK/AU) at the DSP level for incrementality measurement (see incrementality testing for paid media).

Key takeaway: A first CTV test needs a DSP, one 30-second creative, a household frequency cap, and a matched geo holdout, not a media plan.

Creative Requirements That Convert on the Living Room Screen

CTV creative rules are different from social. Sound is on, the screen is 55 inches away, no one is scrolling, and there is no click. Your job is to be memorable enough to lift branded search within 72 hours.

Technical baseline: 16:9 1920x1080 master, 15s and 30s cuts, MP4 H.264 or ProRes, safe title area, minimum 12Mbps bitrate. Subtitles are unreliable on CTV (many apps strip them), so voiceover carries the message. Get brand + benefit + product visual on screen within the first 3 seconds. Land a clear verbal and visual URL/QR code in the last 5 seconds.

Five mistakes that kill CVR on CTV: (1) reusing a talking head Reel with no product visual, (2) sub-USD 300 production quality that reads amateur on a 55-inch panel, (3) missing an on-screen offer, (4) no end frame with the URL, (5) inconsistent brand colour across the 15s and 30s cuts, which breaks recall.

Key takeaway: Living room creative must earn recall in 3 seconds and prompt branded search in 72 hours, or the CTV spend is decoration.

Measuring CTV: Attribution Without a Click

The single biggest reason CTV budgets get killed early is teams measure it with last-click. Last-click undercounts true CTV value by 40 to 60 percent per Prescient AI's 2026 incrementality benchmark and matches what we see on our own client data.

The 2026 measurement stack we run on every CTV account:

UTM discipline is impossible in CTV (no clicks), so stop trying to force it. Treat CTV as a top-of-funnel channel that shows up as branded search, direct, and email/SMS lift 3 to 14 days later.

Key takeaway: Last-click on CTV is a lie, MMM plus geo holdouts plus branded search index is the only honest measurement stack in 2026.

Budget Framework: How Much of Your Paid Budget Should Go to CTV

Under USD 2M ARR, skip CTV entirely and keep pushing on Meta and Google. Between USD 2M and USD 10M, allocate 5 to 8 percent of paid budget to a 90-day CTV test. Above USD 10M, CTV becomes a permanent line item.

Worked example: a USD 5M DTC apparel brand allocates 6 percent of its USD 200K/mo paid budget (USD 12K) to CTV across 90 days: USD 6K Amazon DSP (Prime Video + Fire TV), USD 4K Netflix programmatic via TTD, USD 2K Roku FAST for reach. Expected outcome: 6 to 12 percent branded search lift, 10 to 18 percent blended CAC improvement, MMM validated at day 60. This complements the retail media budget rather than replacing it.

Key takeaway: Under USD 2M ARR skip CTV, from USD 2M to 10M spend 5 to 8 percent of paid budget for 90 days, above USD 10M make it permanent.

Common Mistakes That Burn CTV Budget in the First 30 Days

Most first CTV tests fail the same seven ways. If you avoid these, you keep the budget in market long enough for incrementality to prove itself.

1. Judging CTV on platform or last-click ROAS instead of blended CAC and branded search lift.
2. Skipping frequency caps and paying to hit the same household 12 times per week.
3. Uploading 6s or 60s cuts, which do not match CTV slot lengths and get down-weighted by DSPs.
4. Not excluding the ACR loop, so you retarget the same household with site retargeting immediately after the CTV impression.
5. Buying only Netflix prestige inventory at USD 55 CPM and burning the whole budget on 3M impressions.
6. Ignoring FAST inventory at USD 15 CPM that would double your reach at the same spend.
7. Killing the test at day 14 before MMM or geo holdout has enough data (minimum 45 to 60 days).

Key takeaway: The average failed CTV test dies at day 14 because someone opened a platform ROAS dashboard, not because the channel did not work.

Frequently Asked Questions

What is the minimum budget to test CTV ads in 2026?

USD 3K to USD 5K per month for 90 days is the realistic floor. Below that, you cannot support both a spend cell and a holdout cell, and you will not get statistically meaningful branded search lift. Self-serve platforms like Vibe, MNTN, and Tatari make this budget workable without a trading desk, and Amazon DSP has no self-service minimum since unBoxed 2025.

How is CTV different from OTT?

CTV describes the device (a smart TV, Roku, Fire Stick, Apple TV, or gaming console). OTT describes the delivery method (streaming video over the internet, versus cable or broadcast). All CTV is OTT, but not all OTT is CTV. A Hulu ad on your phone is OTT mobile, the same ad on your living room Roku is CTV. Buying CTV specifically means buying big-screen inventory.

Can small DTC brands buy Netflix ads directly in 2026?

Yes. Since Q1 2026 Netflix programmatic inventory is open on DV360, The Trade Desk, Amazon DSP, and Yahoo DSP with a USD 500/day self-serve minimum. Direct insertion orders still require USD 45K+ commitments and a Netflix ad sales rep. Most DTC brands under USD 25M ARR buy Netflix programmatically inside their existing DSP alongside Amazon Prime Video and Roku.