Most high-consideration brands that want to test CTV without cutting paid search budget are stuck on the same problem. Search is the channel that reports the cleanest return, finance trusts it, and any plan that shrinks it looks like a bet against the one line item that "works." The good news is that a properly built connected TV test does not need a single dollar from search. It needs a separate, bounded budget, a frozen search baseline, and a measurement design that treats search as a dependent variable instead of a funding source. This guide shows how to structure that test, how to fund it, how to keep search steady while it runs, and what results justify scaling.
Why Do Brands Assume a CTV Test Has to Come Out of Search?
Because that is how most agencies pitch it. The typical CTV proposal arrives as a "rebalancing": move 20% to 40% of search spend into streaming, accept a dip in last-click conversions, and wait for brand effects to show up later. For a brand with a five-figure average order value and a 60-day sales cycle, that is an uncomfortable trade. The search conversions disappear immediately and visibly; the CTV benefit, if it comes, arrives slowly and is credited to other channels.
There is also a measurement reason the "rebalance" framing is flawed. If you cut search and add CTV in the same window, you change two variables at once. Any movement in revenue, positive or negative, is impossible to attribute. The CTV test fails not because CTV did not work but because the experiment could not answer the question.
We hear some version of this from prospects: they want to try streaming, but they will not accept an aggressive transition away from search. That instinct is correct. The right answer is not to persuade them otherwise; it is to design a test that never asks them to.
How Do You Test CTV Without Cutting Paid Search Budget?
The principle is simple: CTV is tested as an addition, search is held constant, and search performance becomes one of the outcomes you measure. In practice that takes four decisions.
- Fund the test from a defined, incremental pool. The budget is new or borrowed from a lower-confidence line, never from search.
- Freeze the search baseline. Budgets, bid strategies, targets and keyword sets are locked for the test window, so any change in search performance can be read as a CTV effect.
- Run CTV in some markets and hold it out of matched markets. Geography is the cleanest control available for streaming, because CTV is bought at the household level and almost always reachable by ZIP or DMA.
- Agree the success metrics in advance. Branded search volume, new-customer revenue and incremental cost per acquisition in test versus control markets, judged against thresholds the CFO signs off before launch.
Done this way, the worst-case outcome is a bounded test spend that did not move the business. Search revenue is untouched in every scenario.
Where Should the CTV Test Budget Come From?
If search is off the table, the money has to come from somewhere. These are the four sources we see work, compared on how easy they are to approve and how clean the resulting read is.
| Funding source | Typical size available | Approval difficulty | Effect on measurement | Best fit |
|---|---|---|---|---|
| New incremental test budget | 5% to 10% of quarterly paid media | Moderate; needs a CFO-approved test plan | Cleanest; nothing else changes | Brands with growth targets and finance support |
| Low-confidence display or open-exchange programmatic | Often 10% to 30% of programmatic spend | Low; these lines rarely have defenders | Clean if the cut is applied evenly to test and control markets | Brands running broad prospecting display |
| Paid social prospecting with rising CAC | 10% to 20% of social prospecting | Moderate | Acceptable if social is reduced equally everywhere | Brands where Meta efficiency has declined |
| Sponsorship, print or event budget not yet committed | Varies widely | Low to moderate | Clean; these channels rarely affect the test window | Luxury brands with discretionary brand budgets |
The first option is the gold standard. The second is usually the fastest to approve, because open-exchange display with no audience or supply controls is often the weakest line in a high-consideration plan. If you take money from another channel, cut it evenly across test and control markets, so the only difference between geographies is the presence of CTV.
How Big Does the CTV Test Need to Be?
Big enough to produce a readable signal in the markets you test, and no bigger. For brands selling products and services above $5,000, these are Stillwater planning ranges rather than industry averages:
| Test parameter | Stillwater planning range | Why it matters |
|---|---|---|
| Monthly CTV spend in test markets | $25,000 to $75,000 | Below this, frequency on affluent households stays too low to move behavior |
| Test duration | 8 to 12 weeks, plus a 2- to 4-week post-period | Long consideration cycles need time for search and site behavior to respond |
| Share of addressable households in test markets | 40% to 60% | The rest form the control |
| Weekly frequency per reached household | 2 to 5 exposures | Enough to build memory without waste |
| Premium CTV CPM with affluent targeting | $35 to $70 | Private marketplace deals on premium streaming supply |
If those numbers exceed what you can fund, narrow the geography rather than thinning the spend. A concentrated test in four well-matched markets will tell you more than a diluted national flight. Our longer breakdown of how to size a CTV test budget for high-AOV brands covers the arithmetic for different order values.
How Should Paid Search Behave While the CTV Test Runs?
This is where most tests are quietly ruined. Smart bidding systems react to changes in demand, and if CTV starts generating branded queries, the platform will happily spend more, change bids and shift budget between campaigns. That is good for the business but bad for the experiment, because the search team's actions blur the read.
Lock these elements for the duration of the test:
- Daily budgets on brand and non-brand campaigns, with enough headroom that brand campaigns are not budget-capped in test markets. A budget-limited brand campaign cannot capture new demand, which hides the CTV effect.
- Bid strategies and targets. Do not change target CPA or target ROAS values mid-test.
- Keyword and audience lists. No major additions or exclusions.
- Landing pages and offers. A new offer in week four will swamp any media signal.
- Geographic settings. Search must keep running in both test and control markets on the same terms.
Should You Raise Branded Search Budgets During the Test?
Raise caps, not spend. If brand campaigns were limited by budget before the test, lift the cap everywhere before launch, so that brand search can capture whatever demand exists in every market. Then leave it alone. The spend will rise naturally in test markets if CTV is working, and that rise is part of the evidence.
How Do You Test CTV Without Cutting Paid Search Budget and Still Get a Clean Read?
A geo holdout gives the cleanest answer available to most brands. The design follows a few rules, which we cover in more depth in our guide to geo experiment design for advertising:
- Match markets on history, not intuition. Pair markets by 12 to 24 months of branded search volume, site sessions and revenue, using a synthetic control or matched-market method rather than picking cities that "feel similar."
- Use enough markets. Four to eight test markets and a similar number of controls is a sensible minimum for a regional test; a national brand can use many more DMAs or ZIP clusters.
- Keep everything else equal. Every other channel runs identically in test and control markets.
- Pre-register the read. Write down the metrics, the comparison method and the threshold for success before the first impression serves.
Search is particularly useful here because branded query volume responds quickly to video exposure and is available by geography through the search platforms and Google Trends. Practitioners have long used the comparison of branded search in exposed versus unexposed regions as a practical CTV readout; Search Engine Land's overview of how CTV fuels search demand describes the same approach.
What Should You Measure When You Test CTV Alongside Search?
Measure CTV's effect through search, not in competition with it. These are the metrics that matter, in rough order of how quickly they move:
| Metric | What it tells you | When it typically moves |
|---|---|---|
| Branded search query volume (test vs. control) | Whether CTV is creating active interest | Weeks 2 to 4 |
| Branded search impression share and CPC | Whether new demand is being captured efficiently | Weeks 2 to 4 |
| Direct and organic branded site sessions | Demand that bypasses paid search | Weeks 2 to 6 |
| Non-brand search conversion rate | Whether exposed households convert better when they shop the category | Weeks 4 to 8 |
| Qualified enquiries or consultations | Mid-funnel response for long-cycle products | Weeks 4 to 10 |
| New-customer revenue and incremental CPA | The answer the CFO actually needs | Weeks 8 to 16 |
Two notes. First, do not judge CTV on platform-reported view-through conversions; they credit the channel for buyers who would have converted anyway. Second, expect search to look better in test markets, and record that as a CTV outcome rather than a search win.
What Does a Good Result Look Like?
As Stillwater planning ranges for premium CTV with affluent targeting, a test worth scaling usually shows branded search volume 5% to 20% higher in test markets than in controls, non-brand search conversion rates a few points higher among exposed households, and an incremental CPA within about 1.5 times your blended target, improving over the post-period as long-cycle buyers close.
Vendor studies point the same direction but should be treated cautiously. Lifesight's 2026 CTV measurement report, produced with several CTV sellers, claims a 22.3% lift in paid search conversions from paired CTV exposure, and PPC Land's coverage of the report notes that it discloses no sample or methodology. That is exactly why you run your own holdout: your market, your products, your numbers.
How Do You Present the Test to Your CFO?
Finance teams approve tests that have a ceiling, a timeline and a decision rule. Bring a one-page plan that states:
- The maximum spend and the date it stops, regardless of results.
- The guarantee: paid search budgets and targets do not change.
- The hypothesis, in plain terms: CTV will increase branded search and new-customer revenue in test markets relative to controls.
- The decision rule: what result triggers scaling, what result triggers a second test, and what result ends the program.
- The downside: the full test cost, written as a number, with no hidden dependencies on other channels.
That last point is what wins approval. A test that cannot damage search revenue is a much easier decision than a reallocation that might.
Common Mistakes When You Test CTV Without Cutting Paid Search Budget
- Letting smart bidding run free. Changing search targets mid-test, or allowing automated budget shifts between markets, destroys the comparison.
- Budget-capped brand campaigns. If brand search cannot spend, new demand from CTV goes uncaptured and the test looks flat.
- Too little spend in too many places. A thin national flight produces no readable lift anywhere.
- Buying cheap CTV. Low-cost open-exchange streaming inventory, long-tail apps and unverified supply reach the wrong households and dilute the result. For a high-consideration brand, premium CTV through private marketplace deals is the version worth testing.
- Ending at week six. Long-cycle purchases convert well after exposure. Keep a post-period in the read.
- Crediting search for CTV's work. When branded search improves in test markets, that is CTV's contribution, and it should be reported that way.
- Changing creative or offers mid-flight. One variable at a time.
What Happens After the Test?
If the test clears its threshold, scale gradually. Extend CTV into the control markets first, which turns them into a second confirmation, then expand geography in steps of 25% to 50% while watching incremental CPA. Keep search funded at its pre-test level and let rising brand query volume justify any increase.
If the result is ambiguous, the most common causes are insufficient spend per market, weak creative or poor supply quality. Fix one and retest rather than abandoning the channel. If you are still deciding whether a test makes sense at all, our piece on whether a brand that has never tried CTV should test it is a useful starting point, and our guide to incremental ROAS for CTV explains how to turn test results into a scaling budget.
The point of the exercise is to replace an argument with evidence. When CTV is tested on its own budget against a stable search baseline, the result answers the question every CFO asks: did this add revenue we would not otherwise have had?
Design a CTV Test That Protects Your Search Program
Stillwater Media builds CTV tests for luxury and high-consideration brands that keep paid search intact: premium streaming inventory through private marketplace deals, affluent audience engineering, matched-market geo holdouts and a pre-agreed decision rule. We take on a limited number of engagements each quarter.
If you want to add streaming without putting your search revenue at risk, apply to work with Stillwater Media and we will review whether a test fits your business.
About the author: Stillwater Media is a selective performance media agency for luxury and high-consideration brands, headquartered in Charlotte, NC and working with clients nationally and internationally. The team plans premium CTV, programmatic, affluent audience targeting, DOOH, streaming audio, YouTube Select, podcasts, native and social for brands where customer lifetime value exceeds $5,000 and sales cycles run longer than 30 days, including private aviation, luxury real estate, wealth management, private clubs, luxury automotive, premium DTC and luxury hospitality. Signal. Strategy. Scale.
Frequently Asked Questions
Can you test CTV without cutting paid search budget?
Yes. Fund CTV from a separate, bounded test budget or from a low-confidence line such as open-exchange display, keep paid search budgets and bid targets frozen, and run CTV in matched test markets while holding it out of control markets. Search then becomes an outcome you measure rather than a source of funding.
How much budget does a CTV test need for a high-consideration brand?
As a Stillwater planning range, brands selling products above $5,000 typically need $25,000 to $75,000 a month in CTV spend in the test markets for 8 to 12 weeks, plus a post-period. If that is too much, narrow the geography instead of spreading spend thinly.
Does CTV increase branded search?
Often, yes. Branded search volume is one of the fastest signals of CTV impact and usually moves within two to four weeks in exposed markets. Compare branded query volume in test markets against matched control markets to measure the effect rather than relying on platform view-through conversions.
Should I change my search bids during a CTV test?
No. Lock budgets, bid strategies, targets, keyword lists and landing pages for the test window, and lift any budget caps on brand campaigns before launch. Changing search settings mid-test makes it impossible to separate the CTV effect from search optimization.
How do I know if a CTV test worked?
Agree the decision rule before launch. As Stillwater planning ranges, a test worth scaling usually shows branded search 5% to 20% higher in test markets than controls and an incremental CPA within about 1.5 times the blended target, improving as long-cycle buyers close.
