Why this guide matters now

And what it changes for your CTV strategy

CTV has moved quickly into serious growth planning. Premium inventory is easier to access programmatically, performance teams can measure cross-device outcomes, and advertiser investment continues to rise.

For mobile marketers, the opportunity is broader than attributed installs. CTV can reach audiences outside recent mobile exposure, create demand earlier in the journey, and influence the search, social, app-store, and direct activity that follows.

This playbook brings the practical pieces together. You’ll find regional CPM benchmarks, test-budget guidance, conversion timing, attribution windows, creative best practices, and a 90-day framework for getting started. The aim is to help you understand where CTV fits in your mix, what you should expect from a first campaign, and what evidence you need before scaling it.

What you will learn

1. The state of CTV

For years, CTV sat in the “interesting but not yet” column of most UA budgets. It was expensive, hard to measure, and frankly easier to ignore when Meta and Google were still returning reliable results.

That’s changed. CTV has graduated from experimental to expected. It’s no longer something teams test once and abandon. It’s becoming a standard line item in serious growth budgets, and the teams running it are doing so because it’s working, not because someone told them to diversify.

The market numbers back it up. US CTV ad spend is forecast to surpass linear TV entirely by 2028, reaching nearly $46 billion, according to eMarketer. IAB’s 2026 Outlook Study has CTV growing 13.8% year over year, faster than every major channel except social. Advertiser intent points the same way: in a 2026 survey by Advertiser Perceptions and Premion, 70% of CTV advertisers said they plan to raise their CTV budgets this year, by an average of 17%.

The investment signals point the same direction. According to Nielsen’s 2025 Annual Marketing Report, 56% of global marketers plan to increase investment in OTT and CTV. The IAB forecasts digital video ad spend, including CTV, to exceed $80 billion in 2026, growing faster than the broader ad market.

Demand-side platforms are seeing the same curve. Across the Smadex platform, CTV spend increased by more than 2,000% from 2023 to 2026, with clear acceleration beginning in mid-2025. The full Smadex playbook, including how to plan for a channel in its acceleration phase, is in Section 4.

So what actually changed? Three things, roughly in this order. Attribution caught up, meaning a living-room impression can now be tied to an actual install and downstream ROAS rather than just reach. Supply matured, with inventory becoming available programmatically instead of through private marketplace deals, which let DSPs optimize CTV like any other performance channel. And advertisers stopped treating CTV as a branding line item. Performance marketers now run it on the same KPIs as the rest of their UA mix.

That shift from “CTV as branding” to “CTV as a performance tool” is what this guide is about.

2. What makes CTV different

CTV is not mobile on a bigger screen. The sooner you stop thinking about it that way, the faster your campaigns will improve.

On mobile, you’re fighting for attention. Users are scrolling, swiping, switching apps. Your ad has less than two seconds to stop the thumb. On CTV, the dynamic is completely different. People are sitting back. The screen is the only thing in front of them. And the ads are non-skippable.

The attention environment is real. CTV ads average a 97% completion rate. Not because viewers love ads, but because they can’t skip them and the barrier to leaving the room is high. That’s 30 seconds of uninterrupted exposure on the largest screen in the house.

The ad load is much lower than mobile. Fewer ads per hour means less competition for recall. When you’re one of three ads in a break rather than one of thirty in a feed, memorability works differently.

The conversion path is cross-device by definition. Someone watches your ad on their TV, then picks up their phone and downloads your app. There is no click. There is no direct handoff. The entire journey happens across two devices, with a time delay in between. That’s not a flaw in the system. It’s just how CTV works, and your measurement needs to account for it.

And the second screen is already in the viewer’s hands. 78% of consumers scroll social media while watching TV. The full dual-screen data, and what it means for how you plan cross-screen journeys, is in the tvScientific spotlight on the next spread.

These aren’t reasons to avoid CTV. They’re reasons to approach it differently from every other channel you run.

3. Where CTV fits in your funnel

There’s a question that kills CTV budgets before they get started: “What’s the CPI?”

It’s the wrong question. Not because cost doesn’t matter, but because CTV doesn’t work like mobile display. It doesn’t produce a click. It doesn’t produce an immediate install. What it produces is demand. And that demand shows up in your other channels.

Think of it like this. Performance marketing is picking fruit off trees that already exist. CTV is planting trees. The harvest comes later, and it shows up in your search lift, your organic installs, your social CVR. If you measure CTV by last-click CPI, it will always look bad, because last-click attribution gives CTV zero credit for anything that converted on a different device, at a different time, through a different channel.

So how do you value a channel like that? Not by what it books in isolation, but by what it moves everywhere else. The signals worth watching, branded search, direct app store traffic, conversion rates in exposed geos, all live in other reports, and Section 8 covers how to read them.

The CTV touchpoint funnel

CTV creates the demand that every touchpoint below it captures

There’s a second thing CTV does that no other channel in a typical UA mix can: it reaches genuinely new people. Most mobile networks fish in the same river. The audiences overlap heavily, and adding a fourth or fifth network often just means paying to reach the same users through a different door. CTV reaches audiences at a different touchpoint entirely. Moloco’s data makes this concrete: 94% of users who installed from a CTV ad had seen no prior Moloco ad in the previous 48 hours. The full methodology and what it means for funnel expansion is in the Moloco spotlight in Section 4.

This is also why CTV is not a branding exercise. Framing it that way is the fastest way to lose your budget approval. The goal isn’t reach for its own sake. The goal is building the kind of awareness that makes your performance channels work harder. Those are two different things.

Where CTV earns its place in the funnel

• Reaching audiences that your mobile campaigns have saturated or can’t match
• Creating demand in high-value segments before they enter your retargeting pools
• Supporting campaigns in new geos where brand recognition is low
• Complementing paid social and search for mid-funnel users who’ve seen your ads but haven’t converted

The channel is mid-to-upper funnel. That doesn’t mean it’s soft. It means you need to measure it accordingly.

4. How CTV attribution actually works

This is the section most guides skip. They’ll tell you CTV is measurable without explaining what that actually means, or what the tradeoffs are. There are four main challenges with CTV attribution, and they’re worth naming clearly.

First: it’s cross-device. The ad plays on a TV. The conversion happens on a phone. These are two different devices with no direct connection. There’s no device ID passed from the TV to the mobile app store. Attribution has to infer the link.

Second: it’s impression-only. There’s no click. Mobile attribution is built around clicks. CTV has none. The entire attribution model runs on an impression that happened, at some point, before an install that happened later, on a different device.

Third: conversion timing is unforgiving. Users don’t click; they act on another device, minutes to hours later. Moloco’s data shows 66% of CTV conversions happen within six hours of ad exposure: fast, high-intent action, but far outside the instant feedback loop mobile marketers are used to. And a meaningful tail converts even later. Attribution windows configured for click-based channels miss much of this behavior entirely.

Fourth: CTV networks each handle things differently. Some behave like self-attributing networks. Others support MMP-based attribution. The integration setup varies by partner.

The attribution black hole (and how to avoid it)

If your first CTV report looks disappointing, check your attribution settings before you check your campaigns. There’s a good chance the installs are already there but the credit isn’t.

Here’s why. When several touchpoints could explain the same install, an attribution engine has to pick one, and it picks in order of confidence: a click ahead of a view, a known device identifier ahead of a probabilistic match. That ordering is right for most of your mix. CTV is the exception, because the channel works differently by design. It earns attention on the television while the install happens on the phone, so it produces a view rather than a click, and it matches on the household rather than the device.

Singular accounts for this, and it takes one setting. The “prioritize in attribution decision” setting gives CTV impressions a fair window to compete for credit alongside other signals. The recommended starting point is at least 8 hours. This doesn’t inflate CTV numbers. It removes a structural disadvantage that was deflating them. Before drawing any conclusions about CTV channel ROI, confirm this setting is active. Vibe’s case study in Section 8 shows exactly what happens when a team gets this right.

How Singular’s CTV attribution works

Singular’s CTV attribution uses IP-based household matching. When a CTV ad is served on a smart TV and a mobile app install follows from a device on the same network, Singular connects those two events. This enables marketers to measure CTV’s impact across devices, even when the ad view and install happen on different screens.

The default attribution window for CTV impressions is 24 hours, configurable from 1 to 72 hours depending on how your audience typically behaves. Longer windows capture more delayed conversions; shorter windows are more conservative. CTV touchpoints are the lowest priority in Singular’s attribution stack by default but with a single click, you can elevate CTV’s priority so it’s treated like a deterministic click for a set period, which is the fix described above.

The link setup matters: you need a view-through tracking link (not a click-through link) and a CTV-enabled partner integration. Singular supports CTV integrations with all the top partners like MNTN Performance TV, Moloco, The Trade Desk, tvScientific, Samsung Ads, Smadex, Vibe, AppLovin, Unity CTV, YouAppi, Verve, and others. Additionally, Singular is the first and only MMP recognized in the new Certified Measurement Partner Program from tvScientific by Pinterest.

The methodology stack: what to use and when

Last-click: technically available, practically useless for CTV. Gives the channel almost no credit for anything that didn’t result in a direct, immediate install. Avoid as your primary measurement method.

Priority windows: the practical foundation. You configure your MMP to elevate CTV impression priority in the attribution decision, so CTV gets credit for installs it influenced even when another touchpoint came later. Better than last-click, still limited by the probabilistic nature of the match.

MTA and assists: where CTV’s real story gets told. Multi-touch attribution looks across the whole journey and gives credit in fractions, so a CTV view that came before a paid social click still gets recognized for the part it played. CTV rarely wins the last touch. It sets up the conversion someone else closes, and MTA is what makes that visible.

Incrementality and MMM: the bigger question. These answer ‘what would have happened if we hadn’t spent anything?’ rather than ‘who gets the credit.’ They take longer and cost more, but they’re the closest thing to the truth, and they’re worth running once your CTV budget is meaningful.

The honest take: for most teams, priority windows plus multi-touch attribution is the right starting point. Priority windows make sure CTV is counted fairly. MTA shows you what that spend was actually worth. As the budget grows, incrementality is how you prove it.

5. Creative that works on CTV

CTV creative can not be a resized version of your mobile creatives. The environment is different, the attention dynamic is different, and the conversion path is different. What works is different, too.

The mental shift is this: on paid social, you test for interruption. Can you stop the scroll, win the first second, and get the tap? On CTV, the question changes. Can you hold attention? Can you make the product memorable? Can you create enough intent that the user acts later on another device? Practically, that means optimizing for hold rate rather than hook rate. Catching attention in the first 3 to 5 seconds still matters, but storytelling carries the rest.

Three things CTV creative needs to do well

Communicate quickly. The viewer needs to understand three things fast: what the app does, why it matters, and what to do next. The best CTV creative anchors the product to a recognizable use case or moment. Instead of “this app has routes, alerts, maps, personalization, and notifications,” say “before you drive, know exactly where the weather will hit.” That’s the kind of message someone can remember later when they pick up their phone.

Show the product. On mobile, you can get away with emotion-led creative because the tap-to-download is right there. On CTV, you’re asking someone to remember your app, pick up their phone, find it, and install it. If they’re not completely clear on what it does and why they’d want it, that chain breaks.

Match the environment’s production standard. Creative quality matters in the living room in a way it doesn’t in a feed. That doesn’t mean everything needs Super Bowl production value. Format should align with your brand: UGC-style creative can win decisively (one D2C brand, Hairstory, saw UGC outperform polished creative 4:1), but whatever the format, it has to feel native to lean-back viewing rather than like a resized social ad.

Format guidance

• Start with 30-second spots. Thirty seconds is the CTV standard and gives you enough room to tell a story. Test 15- and 60-second cuts once you have baseline data to compare against.
• Put the brand in the first few seconds, not the last. On CTV, there’s no second chance if someone looks away
• A/B test end card CTAs. The call to action at the end of the ad is one of the highest-leverage creative variables
• Rotate creative every 6-8 weeks at scale. CTV fatigue builds more slowly than social, but it still builds. When you rotate, test different value propositions rather than only swapping visuals
• Test bigger ideas, not micro-variations. The creative testing question on CTV is less “which hook won?” and more “which message created the most incremental demand?”

Beyond the standard video slot

The format landscape is wider than most teams assume, and the line between brand and response formats is blurring. Home Screen Native Ads capture attention before content starts. Pause Ads own the screen during a high-attention moment. Paired with QR codes, these placements create a direct-response bridge from the TV to the phone. The Smadex spotlight in Section 4 covers these formats in more depth.

Response rates: QR codes vs. Roku’s OK button

On QR codes specifically, adoption is real but modest for in-stream ads. Roku’s OK-button ad unit shows what happens when that friction disappears. Viewers respond by pressing OK on their remote: no phone, no extra steps. Peter Hamilton shared on Singular’s Growth Masterminds podcast that response rates on these units reach around 1%, against 0.02% to 0.04% for QR codes. This could eventually translate into conversion rate improvements of 400-500%.

6. Benchmarks

The honest answer on CTV benchmarks is that they vary significantly by vertical, by platform, by format, by region, and by how mature your CTV program is. Numbers that are true for gaming may be off by 2-3x for D2C. Use what follows to size expectations and pressure-test your plan, not to grade your campaigns. A CPM that looks expensive against a benchmark can still buy you the cheapest customers you acquire all quarter.

What the current data supports

• Ad completion rates average around 97% across CTV inventory, one of the strongest engagement metrics in digital advertising
• Test budget floor: most practitioners put the minimum at $25,000/month to accumulate statistically meaningful learnings
• Attribution priority window: at least 8 hours of elevated priority as a starting point, with a 24-hour default view-through window extendable to 72 hours
• Incremental ROAS reference point: a Vibe incrementality test measured 3.03x iROAS, outperforming most of the brand’s other channels (full story in Section 8)
• Conversion timing: 66% of CTV conversions happen within six hours of exposure (Moloco internal data)

7. Getting started

A first CTV test asks for a little more patience than most channels do.

If you come from mobile UA, search, or paid social, you’re used to a fast feedback loop: someone clicked, someone installed, someone bought. CTV works differently. You pay higher CPMs, there’s no click to read, and the early dashboard stays quiet while the channel does its work somewhere you aren’t looking yet.

It’s worth knowing that going in, because the natural reaction is to start hunting for cheaper inventory, and that’s usually where the money actually gets wasted. A first test isn’t about buying cheap impressions. It’s there to answer one question: did quality exposure, to the right audience, create growth you wouldn’t have had otherwise?

Keep the first test clean

One clear audience. One clear outcome. A realistic budget. A measurement plan that compares exposed versus non-exposed behavior. Don’t test six audiences, five creative directions, and three buying strategies simultaneously and then wonder what drove the result.

The question a first test should answer is not “did we get cheap impressions?” It’s “did we buy quality exposure, to the right audience, and did that exposure create incremental growth?” That growth might show up as installs, trials, branded search, app store visits, or stronger performance in other channels. Design the test to detect it.

One practical preparation step: scale your search campaigns before starting the CTV trial. This gives you a stable baseline, so if branded search lifts when CTV goes live, you can read the change with confidence.

And lead with prospecting rather than re-engagement. The reasoning and the measurement bonus that comes with it are covered in the Vibe spotlight in Section 8.

Audience strategy

Start broad. CTV optimization relies on signal density during the learning phase, and restricting publishers, audiences, or contextual inventory before the models establish confidence limits their ability to find incremental opportunities. Let AI-driven optimization establish a learning foundation first, then introduce manual refinements as meaningful performance signals accumulate. YouAppi’s practitioner insights in Section 6 cover this in their own words.

As the program matures, first-party data becomes the durable targeting advantage. Activating CRM records, first-party audiences, app events, and loyalty signals strengthens CTV targeting while reducing dependence on third-party identifiers that are steadily losing reach. Platforms are also opening new intent-signal sources; tvScientific by Pinterest, for example, lets marketers activate Pinterest’s audience signals on CTV.

Partner selection

Don’t test five CTV platforms simultaneously. The more fragmented your strategy, the harder it is to understand what’s working, and frequency capping across fragmented inventory is genuinely difficult. Start with one partner. Learn. Expand.

Look for partners that support MMP-based attribution, offer transparent data access, and can share impression-level data with your MMP. Supply quality matters more than CPM: who are you buying from, where are you actually appearing, and can you understand what was delivered and what happened afterwards?

Campaign setup

Define your goal before you set up the campaign. CPA? Branded search lift? Each implies different bidding, creative, and measurement. The teams that struggle with CTV are usually the ones that didn’t define success before starting.

The most common mistakes, in rough order of damage: measuring CTV with the wrong mental model (judge it like a last-click mobile channel and you’ll undervalue it; judge it like old-school TV and you’ll overvalue it), not setting up incrementality properly, underfunding the test to the point where results are just noise, and resizing a social ad instead of building for the living room.

On bidding: dynamic ML-driven bidding is the standard on CTV platforms. Let the algorithm work and give it time. Moving too fast, pausing or adjusting before attribution cohorts complete, is one of the most common early mistakes.

What good looks like in the first 90 days

CTV rewards patience in a way mobile UA does not. Conversions are delayed, the learning phase is real, and the proof comes from a causal test, not a dashboard on day two. Here is the shape of a disciplined first quarter.

DAYS	1–30	31–60	61–90
PHASE	Launch and learn	Optimize on signal	Prove and decide
What you’re doing	Go broad on targeting and let the algorithm find pockets of performance. Run several genuinely different creative directions, not variations of one. Confirm view-through attribution and priority windows are live before spend starts.	Let differences between audiences and creatives separate, then act on them. Tighten frequency caps into the 3 to 7 per week range as fatigue appears. Scale spend on clear winners at roughly 20 to 30 percent per week, not in one jump.	Run a geo or holdout incrementality test with a pre-set duration. CTV needs 4 to 6 weeks of active flight plus a 2 to 4 week post-treatment window, because a meaningful share of CTV conversions land after exposure ends.
What to watch	Delivery health only: completion rate, frequency, and early CPI. First read-outs land around day 3 to 5.	Segment and creative separation, frequency, and branded-search or app-store lift measured against your pre-launch baseline.	Incremental ROAS and cost per incremental install, read against caused revenue rather than attributed installs alone.
What not to do	Don’t judge ROAS yet, and don’t narrow targeting. You’re buying signal, not outcomes. Cutting early starves the model of the data it needs.	Don’t over-optimize on vanity metrics like impressions or raw completions, and don’t scale everything at once. Chasing week-one noise locks in the wrong winners.	Don’t stop the test early because results look good; that’s the most common way to manufacture a false positive. Decide scale, restructure, or stop on the causal read.

8. Measuring what matters

The right question isn’t “how much revenue did CTV attribute?” It’s “how much revenue did CTV cause?”

Those sound similar. They’re not. Fixing your attribution priority window stops CTV from being written off, but it changes how credit is assigned, not whether revenue was caused. What your MMP attributes to CTV and what a causal experiment reveals are rarely the same number. They measure different things. Drawing conclusions from only one source leads to wrong reads and wrong budget decisions.

The signals worth watching

Building a CTV measurement framework

The practical starting point: use Singular’s advanced CTV attribution for install-level data, with the priority window set the way Section 4 describes. Read CTV through multi-touch rather than last touch, so its assists actually show up. Then, as the budget grows, prove the lift with an incrementality test. Counted, understood, proven, in that order.

CTV won’t always be the channel that gets credit. That’s fine. The goal is growth, not just attribution credit. If CTV is making your other channels more efficient, and you can prove that, you have a defensible budget.

About Singular

Singular is the #1-rated MMP on G2 and the only end-to-end marketing attribution and analytics platform that uncovers true ROI across all marketing channels. We transform complex marketing data into actionable insights by unifying campaign data from thousands of channels with cross-device attribution data. Leading brands like LinkedIn, Nike, WB Games, Airbnb, and Rovio rely on Singular to maximize every marketing dollar, eliminate wasted spend, and drive higher user retention. With superior ROI reporting, advanced fraud prevention, user engagement capabilities, and unmatched data accessibility, marketers can finally drive faster growth with smarter insights.

---

Ready to give Singular a try? Contact us today.