Streaming TV ads have gone from an experiment that only big brands could afford to a channel a mid-size ecommerce store can test with a few thousand dollars. The self-serve platforms got better, the minimums came down, and almost every major streaming service now sells an ad-supported tier. What did not get better is how most advertisers buy it. Teams still plan connected TV like a cheap version of broadcast, then judge it like a search campaign, and conclude that it "doesn't work" after six weeks.
This guide covers where streaming TV inventory actually comes from, what it costs, how to make creative for a screen on the far side of a living room, and how to measure it in a way that survives a skeptical CFO.
What counts as a streaming TV ad
A streaming TV ad is a video ad served into long-form content that someone watches on a television through an internet connection. The industry calls this connected TV (CTV). You will also see OTT ("over the top"), which technically covers any streamed video, including on phones and laptops. When a seller quotes you "OTT" inventory, ask what share of it actually runs on a TV screen. The answer changes the price you should pay.
The places these ads run fall into three groups:
- Premium streaming services with ad tiers. Hulu and Disney+, Netflix's ad plan, Max, Peacock, Paramount+ and Prime Video. Ads run inside professionally produced shows and films, usually in short pods of a few spots.
- Free ad-supported TV (FAST) and AVOD apps. Tubi, Pluto TV, The Roku Channel, Samsung TV Plus and many smaller channels. Huge reach, lower prices, more variable content.
- YouTube on the TV screen. A large and growing share of YouTube watch time happens on televisions. It is bought through Google Ads, which makes it the easiest entry point for teams that already run search, though the creative and measurement rules differ from mobile YouTube. Our guide to in-stream ads on YouTube and Meta covers the mobile side of that inventory in detail.
What unites them is the screen. A streaming TV ad is usually unskippable, plays full-screen with sound on, and is watched from a couch, often by more than one person. That combination is why the channel behaves more like television than like social video, even when you buy it with auction mechanics.
Where to buy streaming TV ads
There are four realistic routes, and the right one depends on your budget and how much control you need.
1. Self-serve CTV platforms. Tools like Amazon DSP's self-serve options, Roku Ads Manager, and specialist CTV platforms aimed at small and mid-size advertisers let you upload a spot, pick audiences and launch with modest minimums. They are the best place for a first test because you can see delivery by publisher and adjust quickly.
2. Direct from the publisher. Buying from Hulu, Peacock or Netflix directly gets you guaranteed placement in specific content, custom integrations, and first-party audience data. It also comes with higher minimum commitments, often in the tens of thousands of dollars, and less flexibility once the flight starts.
3. Programmatic through a DSP. A demand-side platform such as The Trade Desk or DV360 lets you buy across many CTV publishers through private marketplace deals and the open exchange, with one frequency cap across all of it. This is the most efficient route at scale, but it needs someone who knows how to set up deals and audit supply paths. If you are weighing outside help, our breakdown of what a programmatic advertising agency should actually do lists the questions that separate competent partners from resellers.
4. Google Ads for YouTube on TV. You can target TV screens as a device type inside a Video campaign or Demand Gen campaign. It is the fastest way to get CTV impressions live if your team already lives in Google Ads, with the caveat that YouTube on TV is not the same environment as a premium streaming service.
A sensible path for most advertisers: start with one self-serve platform or YouTube on TV, prove the channel moves a metric you care about, then graduate to a DSP or direct deals once you are spending enough that cross-publisher frequency control matters.
What streaming TV ads cost
CTV is sold on a cost-per-thousand-impressions basis, and because nearly every impression is a full-screen, sound-on, unskippable view, CPMs run well above display and social. Typical 2026 ranges look roughly like this:
| Inventory type | Typical CPM range | Notes |
|---|---|---|
| FAST / AVOD apps via self-serve | $15–$30 | Broad reach, mixed content quality |
| YouTube on TV screens | $10–$30 | Auction priced, varies with targeting |
| Programmatic premium (PMP deals) | $25–$45 | Named publishers, better transparency |
| Direct premium streaming | $35–$65+ | Guaranteed placement, highest minimums |
These are planning ranges, not quotes. Holiday quarters, live sports and tight audience targeting all push prices up, sometimes sharply.
The CPM is only half the cost picture. A 30-second spot that looks acceptable on a 65-inch screen usually costs more to produce than a set of social cuts. Many advertisers repurpose existing video, which is fine if it was shot horizontally and holds up at TV resolution.
Before you commit budget, run the math backwards. Put your expected CPM and impression goal into our CPM calculator to see total spend, then ask what that spend needs to return. A $10,000 test at a $30 CPM buys roughly 333,000 impressions. With a frequency of three to five per household, that reaches somewhere between 65,000 and 110,000 households. If your product sells to a narrow audience, that may be plenty. If you sell a $20 impulse item nationally, it will barely register.
Targeting: precise enough to be dangerous
Streaming TV sells itself on precision that broadcast never had. You can target by household demographics, interests, purchase data, geography down to ZIP code, and your own customer lists matched to household identifiers. All of that is real, and all of it can be overused.
The common mistake is targeting so narrowly that you pay a premium CPM to reach people who would have bought anyway. A CTV campaign aimed only at your existing customers will show strong attributed results and very little incremental growth. The channel's real strength is reaching households that match your buyer profile but have never heard of you.
A better structure for a first campaign:
- One prospecting line targeting a lookalike or interest-based audience built from your best customers, with existing customers excluded.
- One geographic split so you can compare markets that received the ads against markets that did not. This is your measurement backbone, covered below.
- A household frequency cap of roughly three to five per week. CTV publishers do not share frequency data with each other, so if you buy across several platforms without a DSP, the same household can see your ad far more often than you intended.
Retargeting site visitors on the TV screen sounds appealing, but the audiences are usually too small to deliver and the TV is a poor place to close a sale someone already started on their phone. Leave retargeting to channels built for it.
Creative for the living room
TV creative breaks most of the rules that work on social. The viewer did not scroll past you; they are sitting through you. That changes what earns attention.
Sound carries the message. Social video is built for silent autoplay with captions. On a TV, the sound is on and the viewer may be looking at their phone. Your voiceover needs to name the brand and the offer clearly enough that someone half-listening still gets it.
Brand early and often. Show the logo and say the name in the first few seconds and again at the end. TV viewers cannot click, so the job is to be remembered when they later search or visit.
Make the call to action searchable. "Search [Brand Name]" or a short, memorable URL works better than a QR code for most audiences, although QR codes do get scanned when they stay on screen long enough and the offer is worth the effort.
Fit the format. Fifteen-second and thirty-second spots are standard. Fifteen seconds is usually enough for an ecommerce brand with a single clear offer. Use thirty seconds when the product needs explaining.
Design for distance. Text that reads fine on a phone disappears on a TV across the room. Use large type, high contrast and fewer words on screen.
If you are running streaming TV mainly to build recognition before other channels convert, the planning principles in our guide to running a brand awareness campaign apply directly: define what "aware" means before you spend, and measure it.
Measuring streaming TV ads honestly
This is where most CTV programs succeed or fail. The platforms will hand you plenty of numbers. Few of them answer the question that matters: did the ads cause sales that would not otherwise have happened?
Completion rate is not a success metric. Because most CTV ads cannot be skipped, completion rates of 95% and higher are normal. A high completion rate tells you the ad played, not that anyone cared.
Attributed conversions overstate impact. Most CTV platforms match households that saw an ad to visits or purchases from devices on the same IP address within a window. This catches real effects, but it also takes credit for people who were going to buy anyway, especially if you targeted existing customers. Treat attributed ROAS as an upper bound.
Geo holdouts tell the truth. The most reliable approach for a mid-size advertiser is a matched-market test. Pick comparable regions, run streaming TV ads in some and hold out the others, then compare the change in total sales, branded search volume and direct site traffic between the two groups. If the test markets lift and the holdouts do not, you have evidence of incrementality that no attribution model can give you.
Watch the halo signals. Branded search, direct traffic and new-customer rate are the leading indicators. Make sure your conversion tracking is clean before the test starts, so a lift in those signals is visible and not lost in tagging errors.
Once you have incremental revenue numbers from the holdout, plug them into the ROAS calculator alongside spend to get a figure you can compare fairly against search and social. If you sell to repeat buyers, compare cost per incremental new customer against their lifetime value using the CAC calculator; a channel that looks expensive on first-order ROAS can be one of your cheapest sources of durable customers.
For brands running several upper-funnel channels at once, a marketing attribution platform with media mix modeling can help apportion credit, but it should be calibrated against the holdout results, not used instead of them.
Streaming TV versus other upper-funnel channels
Streaming TV competes for the same awareness budget as YouTube on mobile, social video and audio. Each has a different job.
- Versus social video: CTV costs more per impression but delivers full attention, sound on, on the largest screen in the house. Social video wins on cost and on direct response; CTV wins on memorability and household reach.
- Versus podcasts: Both build trust through attention, but podcast ads borrow the host's credibility in a way TV spots cannot. If your product benefits from a personal endorsement, compare the approach in our podcast ads guide before shifting budget.
- Versus linear TV: Streaming TV offers better targeting, lower minimums and digital reporting. Linear still reaches older audiences and live events at a scale streaming has not fully matched.
When streaming TV is the wrong buy
Not every business should run CTV. Hold off if:
- Your monthly paid media budget is under roughly $10,000. A meaningful test needs enough spend to reach a measurable share of a market.
- Your search and social campaigns are not yet profitable. Streaming TV amplifies demand; it does not fix a weak offer or a leaky checkout.
- You have no way to measure lift. If you cannot run a geo split or watch branded search, you will be left trusting platform-reported numbers.
- Your audience is extremely narrow, such as a niche B2B buyer. The targeting exists, but the waste at TV CPMs usually makes LinkedIn or account-based approaches more efficient.
A 60-day first test
- Weeks 1–2: Choose two to four test markets and matched holdout markets. Confirm conversion tracking and record baseline branded search, direct traffic and sales.
- Week 2: Produce or adapt one 15-second and one 30-second spot. Check them on an actual TV.
- Weeks 3–8: Run on one self-serve platform or YouTube on TV, prospecting audiences only, frequency capped at three to five per week.
- Weeks 9–10: Compare test and holdout markets. Calculate incremental revenue, incremental ROAS and cost per incremental new customer.
- Decide: Scale, adjust creative or targeting, or stop. All three are valid outcomes of a well-run test.
FAQ
How much do streaming TV ads cost? Most streaming TV ads cost between $15 and $65 per thousand impressions, depending on the publisher and targeting. FAST apps and YouTube on TV sit at the low end; direct buys on premium services cost the most. A useful first test usually needs $10,000 to $20,000 in media.
Can small businesses run streaming TV ads? Yes. Self-serve CTV platforms and YouTube on TV have low or no minimums, and you can target a single city or ZIP codes. The real constraint is not the minimum spend but whether your budget can reach enough households to measure an effect.
Are streaming TV ads skippable? Generally not. Ads on premium streaming services and most FAST channels are non-skippable. YouTube on TV can serve both skippable and non-skippable formats depending on how the campaign is set up.
What is the difference between CTV and OTT advertising? CTV refers specifically to ads shown on a television screen through a smart TV, streaming stick or console. OTT covers streamed video on any device, including phones and laptops. Many sellers use the terms interchangeably, so always ask what share of delivery reaches an actual TV.
