Incremental ROAS Agency for CTV: What to Demand

Any incremental ROAS agency for CTV should be able to answer one question in a single sentence: of the revenue that followed your streaming ads, how much would not have happened without them? Most agencies cannot. They report platform ROAS, view-through conversions, or a blended figure that treats every purchase from an exposed household as caused by the ad. For a luxury or high-ticket brand, where a single closed sale can be worth $5,000 to $500,000 and the buyer was often already researching before the first impression, that number is close to meaningless.

This guide defines incremental ROAS (iROAS) precisely, shows how to calculate it for connected TV, compares the measurement methods an agency can use, gives realistic benchmark ranges, and lists the questions that separate agencies who can prove incrementality from those who only claim to.

What Is Incremental ROAS and How Is It Different From Platform ROAS?

Incremental ROAS is the additional revenue caused by advertising, divided by the advertising spend. "Additional" is the operative word: it counts only the revenue that exceeds what a comparable, unexposed group generated over the same period.

Platform ROAS, by contrast, divides all attributed revenue by spend. Attribution assigns credit to any conversion that follows an exposure inside a lookback window. If your brand already sells to affluent households who search for it by name, many of those buyers would have converted anyway, and the platform still claims them.

The gap between the two is not small. On CTV campaigns aimed at in-market or lookalike audiences, it is common for platform-reported ROAS to run two to five times higher than measured incremental ROAS, and the gap widens the stronger the brand's existing demand. The reason is baseline conversion: the higher your organic and direct volume, the more of it the ad platform can claim without having caused it.

How Do You Calculate Incremental ROAS for CTV?

The formula is simple; the data behind it is what takes discipline.

iROAS = (Revenue in exposed group − Revenue in holdout group, scaled to equal size) ÷ Ad spend

An illustrative example: a private-club brand runs CTV in eight test markets and withholds it from eight matched control markets for 10 weeks. After scaling for population and pre-test trend, the test markets produce $1.9M in membership-application revenue against $1.5M expected from the controls. The lift is $400,000. Media spend in the test markets was $160,000. The incremental ROAS is 2.5.

Three details determine whether that number is trustworthy:

• The revenue definition. For high-consideration brands, use closed-won revenue or qualified pipeline from your CRM, not a site event. A lead form fill is not revenue, and platforms optimizing toward it will find the cheapest form-fillers.
• The measurement window. When the sales cycle exceeds 30 days, a 14-day window cuts off most of the effect. Plan for a test period long enough to capture the cycle, plus a post-period read of four to twelve weeks for late conversions.
• The scaling method. Test and control groups are never identical. Pre-period regression or synthetic control methods adjust for the difference, and the agency should be able to show the pre-test fit.

Which Methods Can an Agency Use to Measure CTV Incrementality?

Not every method produces a defensible iROAS. Ask which one is being used, and why it suits your volume and sales cycle.

For most brands in the $50,000 to $500,000 monthly range with sales cycles longer than a month, geo-based designs are the most credible because they observe real revenue in your own systems rather than a modeled platform conversion. User-level holdouts complement them for faster creative and audience reads. Our guide to geo-experiment design covers market counts, duration, and minimum detectable lift in detail.

Why does "independent" measurement matter?

If the same party sells the media, reports the results, and sets the definition of a conversion, the incentive is to report a flattering number. An agency that runs the test design, the control-group construction, and the readout with your finance team's data, and that welcomes a third-party check, removes that conflict. It is the reason our approach to independent incrementality testing for CTV separates the test design from the media plan.

What Is a Good Incremental ROAS for CTV?

There is no universal benchmark, because a good iROAS depends on your margin and how you value a customer over time. The useful question is whether iROAS clears your breakeven.

Breakeven iROAS = 1 ÷ contribution margin on the first sale, and for high-LTV brands, it can be lower once repeat purchases are included.

A brand with a 40% contribution margin needs an iROAS of 2.5 to break even on the first transaction. A brand with a 60% margin needs 1.67. A private aviation or wealth management firm whose customer generates value over many years may accept an iROAS below 1.0 on the first-year revenue, if the lifetime value math supports it.

As a working range from measured CTV programs in high-consideration categories, incremental ROAS on prospecting campaigns commonly lands between 0.8 and 3.0 on first-purchase revenue, with the low end typical of categories with very long cycles and the high end of shorter-cycle premium DTC. Results above 4.0 in a first test should be checked for contamination, such as overlap with a concurrent promotion or a control group that was disturbed by other media.

These ranges are planning aids, not promises. What matters is that the agency states the breakeven in advance and reports against it.

What Should You Ask an Incremental ROAS Agency Before You Hire Them?

Use these questions in your selection process. An agency with real experimental capability will answer each specifically, with examples.

• What method will you use for our first test, and why? A credible answer names a design, the number of markets or the holdout share, and the expected duration.
• Who builds and owns the control group? It should be documented, and you should be able to inspect it.
• What revenue source will the result be based on? The answer should be your CRM, POS, or finance system, not a pixel.
• What minimum detectable lift do we have at our budget? If the agency cannot compute this before launch, the test may be underpowered and inconclusive by design.
• How do you handle overlapping campaigns? Search, paid social, email promotions, and PR moments can all contaminate a test.
• Will you report a confidence interval? A single iROAS number without an interval hides the uncertainty. A result of 2.1 with a range of 0.6 to 3.6 is a very different finding from 2.1 with a range of 1.8 to 2.4.
• What do you do with a null result? A good agency treats "no measurable lift" as an answer that changes the plan, not as a failure to be explained away.
• Can we see the raw, market-level data? If not, you cannot verify the calculation.

What Are the Most Common Mistakes in CTV Incrementality Reporting?

Mistakes in this area tend to be systematic, and most flatter the media.

• Reporting view-through conversions as incremental. A view-through window counts exposed households who later converted. Without a control group, that is attribution, not incrementality.
• Testing too briefly. Ten days of data on a 60-day sales cycle measures only the fastest buyers and understates the effect on high-consideration purchases.
• Underpowered designs. Running a test in four markets with a few dozen weekly conversions cannot detect a 10% lift. The result will say "no significant effect," which is not the same as "no effect."
• Changing the plan mid-test. Shifting budget, creative, or audiences during the test window breaks the comparison.
• Ignoring spillover. Streaming audiences cross market lines, and travel-heavy affluent households may be exposed in one region and convert in another. Buffer zones and market pairing reduce this.
• Measuring only the last quarter of the funnel. CTV often works upstream, raising branded search and direct traffic. An iROAS built only on the ad-attributed conversion misses that halo, while one built on total revenue in the region catches it.

How Should Incremental ROAS Change Your Budget Decisions?

Once you have a credible iROAS, it should govern three decisions.

Where to scale. Compare incremental ROAS by channel and by audience. If CTV shows 1.9 while paid social prospecting shows 0.9, reallocating a portion of the social budget is defensible, subject to the caveat that returns diminish as spend rises.

How far to scale. Average iROAS from a test at one spend level does not tell you the return on the next dollar. Marginal ROAS falls as you saturate the audience. A well-run program tests two or three spend levels, or steps spend up in staged increments across markets, to find the point where marginal returns cross your breakeven.

What to fix before spending more. A low iROAS is diagnostic. It may indicate weak creative, an audience that overlaps too heavily with existing customers, frequency set too low to register, or a conversion path that loses buyers after the ad. Each has a different remedy, and the test design should let you see which.

We also connect results to incremental cost per acquisition, since for lead-based businesses iCPA and iROAS are two views of the same measurement and should reconcile.

How Does Stillwater Media Run an Incremental ROAS Engagement?

The process we use with luxury and high-consideration brands follows a fixed order, because skipping steps is where most tests fail.

• Define the revenue truth. We agree on the revenue source, the conversion definition, and the sales-cycle window with your finance or CRM owner before any media runs.
• Compute breakeven iROAS. Margin and lifetime value set the threshold the result will be judged against.
• Power the test. We calculate minimum detectable lift for your volume and budget, and adjust market count, duration, or spend so the test can produce a readable answer.
• Build the control. Matched or synthetic controls are constructed from pre-period data, with the pre-test fit documented.
• Launch on private marketplace inventory. Running through private marketplace deals on premium streaming supply keeps the variable under test to the advertising itself rather than to inventory quality.
• Hold the plan constant. No mid-test changes, and a log of any outside events such as promotions or press.
• Read out with intervals. You receive the iROAS, its confidence range, the incremental revenue and its source, and a recommendation tied to your breakeven.
• Feed the result back. The finding calibrates ongoing optimization and any marketing mix model, so day-to-day decisions reflect measured value rather than platform credit.

This approach also means we may tell you that CTV is not working for your brand at the current audience or frequency. That finding is worth more than a flattering report, because it protects the budget.

What Should You Do Next?

If your current agency reports ROAS without a control group, request a single incrementality test on your largest prospecting channel this quarter. If the agency resists, cannot scope the test, or offers only view-through numbers, you have learned something about how it measures its own work.

Stillwater Media works with a limited number of luxury and high-consideration brands each quarter, and every engagement starts with a measurement plan before a media plan. If you would like an incremental ROAS framework built around your margins, sales cycle, and CRM data, apply to work with Stillwater Media.

About Stillwater Media. Stillwater Media is a selective performance media agency for luxury and high-consideration brands, serving clients across the US and internationally and headquartered in Charlotte, NC. We specialize in premium CTV on Disney+, Netflix, and Prime Video, programmatic advertising, and affluent audience engineering, with private marketplace access, brand safety controls, and incrementality testing built into every engagement. Signal. Strategy. Scale.

Frequently Asked Questions

What is incremental ROAS in CTV advertising?

Incremental ROAS is the additional revenue caused by CTV advertising, measured against a control group that did not see the ads, divided by ad spend. Unlike platform ROAS, it excludes purchases that would have happened anyway, which makes it the right figure for budget decisions.

How do you measure incrementality for connected TV?

The most credible methods are geo holdout tests, matched-market or synthetic-control designs, and household-level holdouts. The best choice depends on your conversion volume, number of markets, and sales cycle, and the revenue should come from your own CRM or finance data.

What is a good incremental ROAS for CTV?

A good result is one that exceeds your breakeven iROAS, which equals one divided by your contribution margin. As a planning range, prospecting CTV in high-consideration categories often measures between 0.8 and 3.0 on first-purchase revenue, with lifetime value justifying lower figures for high-LTV brands.

Why is platform-reported ROAS higher than incremental ROAS?

Platform ROAS credits every conversion that follows an exposure, including buyers who were already going to convert. The stronger your existing demand, the larger the gap, and it is common for platform figures to run two to five times higher than measured incremental results.

What should I ask an agency about incrementality before hiring?

Ask which test design it will use, who builds the control group, what revenue source the result rests on, what minimum detectable lift your budget supports, and whether it reports confidence intervals. An agency that cannot answer these specifically is probably reporting attribution rather than incrementality.