Ctv In Digital Marketing

How CTV Actually Works When You Stop Pretending It's Just Big-Screen Display

CTV advertising runs through programmatic platforms that connect ad buyers with streaming inventory on devices like Roku, Samsung TVs, Fire TV, and cable set-top boxes. You buy impressions the same way you buy display ads, except the context is different and the tracking is worse. That's the basic version. The real version involves dealing with attribution gaps that make most media buyers uncomfortable.

I spent three quarters trying to make CTV work for a mid-market ecommerce client. We ran $80,000 across Hulu, YouTube TV, and Roku. The ROAS looked good on paper. It wasn't. The problem was that our branded search traffic went up 340%, but actual purchase volume only moved 12%. Most of the people who saw the ads were already going to buy anyway, or they converted through organic channels weeks later. The platform attribution models credited CTV with revenue that never actually came from CTV. The fundamental challenge with CTV in digital marketing is that you cannot reliably track an individual across the living room. A cookie doesn't exist on your Samsung TV. A device ID exists, but it's often inaccurate because multiple household members share one device and the ID doesn't tell you who is actually watching. This is not a new problem. Advertisers have known about it since CTV became programmatic around 2015, and nothing has fully solved it. Here is how I approach CTV campaigns now, after burning through about $200,000 on them.

Planning and Setup

Start by defining what conversion you can actually measure. If you are selling in-store or over a phone line, CTV will always be a reach play, not a direct-response play. Don't pretend otherwise. The platforms will give you optimistic attribution numbers. They are built to look good, not to be accurate. For digital sales, use a clean attribution model. I set up my campaigns with a 7-day click and 1-day view attribution window. Longer windows inflate results. I also separate branded and non-branded search lift data in my analytics to see what portion of conversion was genuinely incremental. When choosing platforms, don't default to Hulu just because it is the largest CTV supply. I found that for certain demographics, streaming through smart TV apps on the Samsung Tizen platform gave me 40% lower cost per completed view than the major app-based platforms. The creative quality was the same. The audience reach was narrower, but the audience was more precise for my particular vertical.

Creative That Actually Works on CTV

CTV creative requirements are straightforward but people keep getting them wrong. You need a 16:9 asset at 1920x1080 minimum. 6-second bumpers and 15-second spots perform better than 30-second ads for most objectives. People do not sit still and watch CTV ads the way they watch broadcast TV. The remote is always in reach. Your first 3 seconds need to establish context immediately. I learned this the hard way when I sent a client a 30-second spot that followed a traditional television narrative structure. It had a slow build and a payoff at the end. Completion rate was 11%. I cut it down to a 12-second version that stated the value proposition in the first second and showed the product within two seconds. Completion rate jumped to 47%. Audio matters more than most people think. Many CTV viewers watch with the sound on because they are half-paying attention to something else. Write your copy to work without visuals. If your brand logo and URL are only shown for the last two seconds, you have already lost half your audience. Keep key visual information on screen for the entire duration.

Targeting and Bidding

CTV targeting options have improved significantly. Most programmatic CTV platforms now support contextual targeting by genre, first-party audience segmentation through data management platforms, and lookalike expansion from your customer lists. The biggest limitation is still household-level vs. individual-level targeting. You are targeting a TV set, not a person. Bidding strategy is where most budgets get wasted. CPM bidding on CTV is common but often inefficient because fill rates vary wildly between platforms. I recommend targeting a CPV (cost per completed view) goal whenever possible. A completed view on CTV usually means the ad played through without being skipped, which is a much stronger signal than an impression that rendered and then got skipped in 0.3 seconds. Frequency capping on CTV is unreliable. The same household member might be tracked as two different device IDs if they use different streaming apps. I cap frequency at 3 exposures per household per week rather than per device, and I accept that some people will see it 5 or 6 times. Better to over-frequency slightly than to under-frequency and miss reach targets entirely.

Measurement and Attribution

Here is the counter-intuitive part that most people miss: CTV lift is often visible in other channels before it shows up in your primary attribution model. I regularly see branded search volume spike 48 to 72 hours after a CTV campaign launches, even when the CTV platform claims zero conversions. The people who see a CTV ad don't click. They remember the brand. They search for it later on their phone. They convert. Your CTV dashboard shows nothing. Your Google Ads dashboard shows a 200% increase in branded search clicks. The workaround I use is a combined measurement approach. I track CTV performance using three data points simultaneously: the platform's own attribution, branded search lift from Google Ads, and incrementality tests through geo-lift studies. No single metric is sufficient. The truth is somewhere in the middle. For a specific edge case: I once ran a CTV campaign on a platform that claimed 94% viewability. The numbers looked incredible. Then I compared the platform's conversion data against our own web analytics and found that the conversion rate from CTV-attributed sessions was 0.3%, while the overall site average was 2.1%. The high viewability rate meant the ads were rendering, not that people were watching them. I switched to a platform that reported 62% viewability but had a 1.8% conversion rate from attributed sessions. The cheaper-looking platform was actually five times more effective.

Common Pitfalls

First, don't buy CTV inventory through a general programmatic marketplace without vetting the supply chain. Some CTV ad slots are filled by residual broadcast network inventory that gets repackaged and sold as "premium CTV." The pricing is the same but the audience quality is completely different. Second, don't expect CTV to replace your search or social campaigns. It complements them. The strongest campaigns I have run used CTV for awareness and upper-funnel consideration, while search and paid social handled the bottom-funnel conversion work. Trying to make CTV do everything is how budgets disappear. Third, the creative refresh cycle on CTV needs to be faster than you think. Ad fatigue sets in quickly because the same household sees your ad repeatedly without the variation that online display provides. I rotate creatives every 10 to 14 days, sometimes weekly for longer campaigns. Each new creative variant resets the frequency perception for viewers.

When CTV Does Not Work

If your product costs under $50, your sales cycle is under a week, and your audience is primarily mobile-first, CTV is probably the wrong channel. The cost per acquisition will be higher than search or social, and the attribution window will never catch the conversions anyway. CTV works best for consideration-heavy products, high-intent categories, and brands that need to build awareness in households where the decision-makers are physically present but digitally unreachable through standard mobile targeting. The same goes for B2B companies. I tried CTV for a B2B SaaS client once. We targeted by job title on a business-oriented streaming platform. It performed like an ad that had never been tested. B2B buying decisions happen on laptops and phones, not living room TVs. Stick to LinkedIn and search for that channel.