Streaming Ads Are Not TV Ads. Stop Running Them That Way.

Paid advertising on streaming and connected TV sits in an awkward middle ground for most marketing teams. It looks like television, gets bought like digital, and gets evaluated like neither. The result is a channel that is chronically misunderstood, routinely misused, and consistently underperforming its actual potential.

CTV and streaming inventory gives advertisers something broadcast television never could: addressable audiences, measurable completions, and the ability to test creative at a fraction of traditional production cost. But those advantages only materialise if you treat the channel on its own terms, not as a cheaper version of a linear TV buy or a longer version of a pre-roll ad.

What Actually Makes CTV Different From Other Paid Channels

When I was at iProspect, we spent a lot of time thinking about channel architecture: which channels create demand, which channels capture it, and how they connect. Most of what we managed was firmly in the demand capture category, paid search being the obvious example. CTV sits at the other end of that spectrum, and that distinction matters more than most teams acknowledge when they start buying it.

Connected TV refers to any television set connected to the internet, whether through a smart TV operating system, a streaming device like a Roku or Fire TV, or a gaming console. Streaming, in this context, means the ad-supported content delivered through those devices, including services like Hulu, Peacock, Pluto TV, Tubi, and the ad-supported tiers of platforms like Disney+ and Netflix. The inventory is bought programmatically, often through DSPs, but it is consumed in a lean-back environment on a large screen with household-level viewing behaviour.

That combination creates a genuinely unusual advertising context. You have the targeting precision of digital, the screen size and attention environment of traditional television, and the measurement infrastructure of programmatic. In theory, that is a powerful combination. In practice, most teams either default to repurposing broadcast TV creative or treat it like a video extension of their paid social strategy. Both approaches miss the point.

If you want to understand how CTV fits into a broader paid advertising strategy, the paid advertising hub covers the full channel landscape, from search and social to programmatic and beyond.

How Streaming Inventory Is Actually Bought and Priced

CTV inventory is predominantly sold on a CPM basis, with prices that vary significantly depending on the platform, the content category, the targeting parameters applied, and whether you are buying programmatically or through a direct deal. Premium publisher inventory on major streaming platforms commands CPMs that would make a paid search manager wince. Programmatic open exchange CTV inventory is cheaper but comes with quality and brand safety considerations that require active management.

The buying landscape breaks down into a few distinct routes. Programmatic guaranteed deals give you reserved inventory at a fixed CPM. Private marketplace (PMP) deals offer curated inventory from specific publishers with some negotiated terms. Open auction programmatic gives you the broadest reach at the lowest floor price but the least control. Direct buys with individual streaming platforms sit outside the programmatic stack entirely and are typically only viable at meaningful spend levels.

One thing worth understanding clearly: the CTV supply chain is fragmented and not always transparent. There are intermediaries between DSPs and publishers, and the reported inventory quality does not always match what is actually delivered. This is not unique to CTV, but it is particularly pronounced in a channel that is growing faster than its measurement infrastructure. Running through a reputable DSP with strong supply path optimisation (SPO) practices is not optional, it is basic hygiene.

For context on how programmatic pricing and quality dynamics play out across digital channels more broadly, this piece from Search Engine Journal on smart pricing mechanics is worth reading as background on how quality signals affect cost in ad auctions generally.

The Completion Rate Trap

I have sat in enough post-campaign reviews to know that completion rate is the metric that gets cited most often when a CTV campaign is being defended. “We hit a 97% completion rate” is presented as evidence of success. It is not. Non-skippable ad placements, which account for the majority of CTV inventory, will generate high completion rates regardless of whether the creative is any good. The viewer finished watching your ad because they had no alternative, not because it held their attention.

This matters because optimising toward completion rate as a primary KPI creates a false sense of performance. The metrics that actually tell you something about whether a CTV campaign is working are harder to collect and less flattering. Brand lift studies, measured through third-party providers, give you attitudinal data. Household-level conversion analysis, where you match CTV-exposed households against conversion events, gives you some directional signal on downstream impact. Incrementality testing, where you hold out a matched audience segment and compare outcomes, is the most rigorous approach but requires scale and planning discipline that most teams do not have in place.

The honest position is that CTV attribution is imperfect. It contributes to outcomes it does not get credited for, particularly in the consideration and intent stages of a purchase experience. That does not mean you should not invest in it. It means you should be honest about what you can and cannot measure, and build your budget case on a realistic assessment of its role rather than metrics that look good in a slide deck.

Frequency Management Is the Biggest Operational Problem Nobody Talks About

Here is a problem that plays out consistently across CTV campaigns at every budget level. A viewer subscribes to four or five ad-supported streaming services. They spend an evening watching content across those platforms. Without cross-platform frequency capping, they can see the same ad from the same advertiser ten or fifteen times in a single session. Not because the media buyer intended that outcome, but because each platform applies its own frequency cap independently, with no visibility into what is happening on competing platforms.

This is not a minor inconvenience. Excessive frequency is actively damaging. It erodes brand perception, it wastes budget, and it creates exactly the kind of advertising experience that makes people resent the brands running the ads. I have never met a marketer who intended to show their ad fifteen times to the same household in a week. But I have seen plenty of campaigns where that was the operational reality.

Managing this properly requires either buying through a single DSP that has cross-platform reach and applies unified frequency caps, or accepting that you will have some degree of over-exposure and building that into your campaign structure. Some DSPs have made meaningful progress on cross-platform identity resolution for frequency management purposes. It is worth asking your buying partner specifically how they handle this, rather than assuming it is being managed.

The same discipline applies to creative rotation. Running a single creative asset at scale across a multi-week campaign will exhaust even a genuinely good piece of work. CTV campaigns need a minimum of two or three creative variants in rotation, with clear rules for when each is served and how you evaluate which is performing.

Audience Targeting on CTV: What Is Actually Available

The targeting capabilities on CTV are more sophisticated than most marketers realise and more limited than the sales decks suggest. What is genuinely available varies significantly by platform and buying method.

First-party data from streaming platforms, including viewing behaviour, content preferences, and subscription data, is available through direct buys and walled garden PMPs. This is valuable because it is based on actual behaviour within the platform, not modelled or inferred data. Hulu’s first-party audience segments, for example, are built on real viewing patterns, not demographic proxies.

Programmatic CTV buying through DSPs typically relies on third-party data segments, household-level identity graphs, and contextual signals. The quality of these segments varies considerably. Household-level targeting is the standard unit in CTV (as opposed to individual-level targeting in mobile and desktop), which means your targeting is inherently less precise than what you are used to in paid social. A household segment built around “sports enthusiasts” might accurately reflect the primary viewer but completely miss the household member who is actually your target customer.

Retargeting on CTV is possible but operationally complex. Matching your CRM data or website visitor lists against CTV household identity graphs works reasonably well at scale, but the match rates are often lower than teams expect, and the methodology varies by data provider. It is worth treating CTV retargeting as a complementary tactic rather than a precision tool.

For teams building out coordinated paid strategies that include social and streaming, Semrush’s overview of paid social strategy is a useful reference point for how audience segmentation principles translate across different paid channels.

Creative Requirements That Most Briefs Get Wrong

I judged the Effie Awards for several years, and one thing that becomes clear when you review hundreds of campaigns is how rarely creative is actually built for the environment where it will run. CTV creative is a specific discipline, and it sits between broadcast television and digital video in ways that require deliberate choices rather than compromises.

A 30-second broadcast TV spot repurposed for CTV is not inherently wrong, but it often underperforms because broadcast creative is built around the assumption of distracted, passive viewing with a remote control nearby. CTV viewing is more intentional than traditional broadcast (the viewer chose to open a streaming app and play a specific show) but less interactive than mobile or desktop. The viewer is not going to click through. They are not going to search your brand name mid-ad. The creative needs to do its work entirely within the window of the placement.

Practically, this means a few things. Brand identification needs to come earlier than in a traditional TV spot. The first five seconds matter more than they do in a skippable pre-roll context, not because the viewer will skip (they cannot) but because attention is highest at the start of a non-skippable placement. The call to action needs to be memorable rather than clickable: a URL, a search term, or a brand name that the viewer can act on later rather than immediately.

There is also a strong case for producing CTV-specific creative at shorter lengths. Fifteen-second non-skippable placements are widely available and often more efficient than 30-second spots when the creative is built specifically for that format rather than cut down from longer work.

How CTV Fits Into a Coordinated Paid Strategy

The mistake I see most often with CTV is treating it as a standalone channel with its own budget, its own KPIs, and no structural connection to the rest of the paid media strategy. That approach produces campaigns that are hard to justify and easy to cut, because they appear to exist in isolation from the outcomes the business actually cares about.

CTV works better as a coordinated layer in a broader paid architecture. At its most straightforward, this means using CTV to build awareness and consideration among audiences who are then retargeted through paid social and paid search as they move through the purchase experience. The CTV exposure creates familiarity and intent. The downstream channels capture and convert it. Neither works as well without the other, but the measurement usually only credits the last touch.

When I was working on large retail accounts, we ran a version of this model where CTV was used to drive search query volume for specific product categories. The logic was straightforward: if we could get a household to see a compelling product ad on their TV in the evening, some percentage of them would search for it within the next 24 to 48 hours. We tracked branded and category search volume as a leading indicator of CTV effectiveness, which gave us a more honest signal than completion rates alone.

Influencer-driven content is increasingly being packaged into paid media strategies that include streaming placements. Later’s guide on influencer marketing and paid media covers how these content types are being distributed across paid channels, which is relevant context if you are thinking about how creator content performs in a CTV environment.

For paid social specifically, the audience data and reporting infrastructure available through platforms like Sprout Social’s paid social analytics can help you understand how CTV-exposed audiences behave differently in social environments, which is useful for building the cross-channel case internally.

The organic versus paid dynamic is also worth considering. Buffer’s analysis of organic and paid social strategy makes a point that applies equally to CTV: paid amplification works best when it is reinforcing something that already has genuine audience resonance, rather than trying to manufacture attention from scratch.

Budget Thresholds and When CTV Makes Commercial Sense

CTV is not a channel for every budget. The CPMs are high relative to display and paid social, the creative production requirements are more demanding than static or short-form social ads, and the measurement infrastructure requires investment to do properly. Below a certain spend threshold, you are unlikely to generate the reach needed to produce measurable outcomes, and the cost of doing it properly (creative, measurement, DSP fees) will consume a disproportionate share of the budget.

There is no universal threshold, because it depends on your category, your target audience size, and your geographic focus. But as a directional guide, CTV starts to make commercial sense when you have enough budget to sustain meaningful reach against a defined audience over a campaign period of at least four weeks. Anything shorter makes it very difficult to build the frequency needed to generate measurable brand impact, and anything below the reach threshold means you are paying premium CPMs for negligible coverage.

For smaller budgets, the ad-supported tiers of major streaming platforms sometimes offer self-serve buying options that lower the entry point. These are worth exploring if you want to test the channel before committing to a programmatic buy. The targeting and measurement capabilities are more limited, but the operational simplicity makes it a reasonable starting point.

Geographic targeting is one area where CTV has a genuine advantage over traditional broadcast. The evolution of regional targeting in digital advertising has been significant, and CTV has inherited that infrastructure. You can run CTV campaigns targeted to specific DMAs, cities, or even zip codes, which makes it viable for regional businesses that would never have considered television advertising under the traditional broadcast model.

Measurement: What Honest Looks Like

There is a version of CTV measurement that looks rigorous but is largely theatre. Running a pixel on a post-view conversion window and attributing all conversions that happen within seven days of a CTV impression to the CTV campaign is not measurement. It is post-rationalisation. The view-through attribution model, which is the default in many DSP reporting interfaces, will almost always make CTV look more effective than it is, because it credits the channel for conversions that would have happened anyway.

Honest CTV measurement starts with defining what you are trying to move and how you will know if it moved. For brand campaigns, that means pre and post brand lift measurement, tracking awareness, consideration, and purchase intent among exposed versus unexposed audiences. For performance-oriented campaigns, it means incrementality testing: holding out a matched audience segment, running your campaign against the exposed group, and comparing conversion rates between the two groups over the campaign period.

Neither approach is perfect. Brand lift studies have methodological limitations. Incrementality tests require scale and clean experimental design. But both are more honest than view-through attribution, and both give you information you can actually use to make better decisions about the channel.

The broader paid advertising landscape rewards teams that are honest about what each channel can and cannot measure. If you are building out your paid strategy and want a framework for thinking about channel measurement across the full mix, the paid advertising section of The Marketing Juice covers attribution, channel selection, and measurement frameworks in more depth.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions