If you are working out how to choose a CTV advertising agency, judge candidates on five things, in this order: where the inventory comes from, who the audience data says you are reaching, how results will be measured, what you pay in total, and who owns the accounts and data when the contract ends. Creative, dashboards, and case-study logos matter less than those five. The 12 questions below turn each one into something an agency has to answer in writing, and the scorecard at the end lets you compare proposals side by side.
We are a CTV agency, so read this with that in mind. We have written it so the questions apply to us as much as to anyone else, and we would encourage you to send the same list to every firm on your shortlist.
What Does a CTV Advertising Agency Do?
A connected TV agency plans, buys, and measures advertising that runs inside streaming content on television screens. In practice the work breaks into five jobs:
- Supply. Securing inventory on services such as Disney+, Hulu, Netflix, Prime Video, Peacock, Max, and Paramount+, through direct deals, private marketplaces (PMPs), or the open exchange.
- Audience. Building target segments from first-party data, third-party data, and automatic content recognition (ACR) data, and matching them to households.
- Activation. Operating a demand-side platform (DSP), such as The Trade Desk, DV360, Amazon DSP, or StackAdapt, to bid, pace, and cap frequency.
- Verification. Confirming ads ran on real TVs, in real content, in front of real households.
- Measurement. Connecting exposure to business outcomes.
Agencies differ most in jobs one and five. Almost anyone can open a DSP seat and launch a campaign. Far fewer can document their supply paths or design a valid lift test.
How to Choose a CTV Advertising Agency: The 12 Questions
Supply and inventory
1. What share of our impressions will run on named, premium streaming services, and through which deal types?
Ask for a projected breakdown by publisher and by deal type (programmatic guaranteed, PMP, open exchange). A strong answer names services and commits to a floor, for example 80 percent or more on a defined list. A weak answer cites "access to 200+ premium publishers" without percentages. Long-tail apps and free ad-supported channels have a place in some plans, but they should be a choice, not a default. Our overview of programmatic deal types explains the trade-offs.
2. Will we receive app-level or publisher-level delivery reports?
Transparency in CTV reporting is uneven, because some publishers pass limited content signals. You should still receive spend and impressions by publisher or app bundle. An agency that reports only "CTV" as a single line is asking you to trust what you cannot inspect.
3. What percentage of delivery reaches a television screen?
"CTV" line items sometimes include mobile and desktop video. Ask for a device-type breakdown and a contractual minimum, typically 90 to 95 percent big-screen delivery for a campaign sold as CTV.
Audience and data
4. Where does the audience data come from, and what is the household match rate?
For an affluent or high-net-worth target, ask which providers supply wealth, income, and asset signals, how segments are defined (a $150K income threshold is a different audience from $2M in investable assets), and what share of the segment matches to addressable streaming households. Match rates in the range of 40 to 70 percent are common; a claim of near 100 percent deserves scrutiny. See affluent audience targeting for how segments are built.
5. How will frequency be controlled across publishers and devices?
Frequency is the most common source of waste in CTV. Without a cross-publisher cap, the same household can see a spot 20 or more times a week while others see none. Ask what identity graph is used, what the proposed cap is (for most high-consideration brands, 2 to 4 exposures a week), and for a frequency distribution report, not just an average.
Measurement
6. How will you prove the campaign caused sales?
This is the question that separates agencies most clearly. Acceptable answers describe a geo-holdout or matched-market test, a household-level holdout, or a third-party lift study, with outcomes read from your own sales or CRM data. An answer built only on view-through attribution, where any household that saw an ad and later visited the site counts as a conversion, overstates results because many of those households would have visited anyway. We cover the design in independent incrementality testing for CTV.
7. Which third parties verify delivery and measure outcomes?
Look for named vendors: DoubleVerify, Integral Ad Science, or HUMAN for invalid traffic and brand suitability; iSpot, VideoAmp, Nielsen, or an incrementality platform for outcomes. An agency that grades its own work with its own pixel has a conflict of interest, however honest the team.
8. What metrics will you report, and which will you optimize toward?
Video completion rate on CTV is typically 95 percent or higher because ads are non-skippable, so it tells you little. Useful reporting includes unique household reach, frequency distribution, cost per incremental visit or lead, and incremental cost per acquisition. For long sales cycles, ask which leading indicator the agency proposes and how it was validated against closed business.
Commercial terms
9. What is the total cost between our dollar and the publisher?
Request a written fee waterfall: agency fee, DSP fee, data fees, verification fees, and any supply-side or curation fees the agency can see. Ask directly whether the agency buys inventory as a principal and resells it at a margin. Principal-based buying is legal and sometimes cheaper, but it must be disclosed, because it changes whose interest the agency is serving. Our piece on programmatic fee transparency lists the audit rights to request.
10. Who owns the DSP seat, the data, and the audiences?
If the agency owns the seat, learn what you take with you when you leave: log-level data, audience definitions, creative performance history, and PMP deal IDs. At minimum, the contract should guarantee data export and a transition period.
11. What are the minimum spend, the term, and the exit clause?
Common structures are a three-month initial term followed by month-to-month with 30 days' notice. Twelve-month lock-ins with no performance exit favor the agency. Minimums tell you who the agency is built for: a firm with a $10K monthly minimum and one with a $75K minimum are running different businesses.
Team and fit
12. Who will work on the account, and how many accounts do they carry?
Ask to meet the strategist and the trader, not only the sales lead. A trader managing 25 accounts cannot review your placements weekly. Ask also for a reference in a category with a similar price point and sales cycle; experience selling $40 subscriptions does not transfer cleanly to $40,000 purchases.
Which Fee Model Is Fairest?
There is no single correct model, but each creates a different incentive.
| Fee model | How it works | Typical range | Incentive to watch |
|---|---|---|---|
| Percentage of media | Agency takes a share of spend | 10 to 20 percent | Agency earns more when you spend more |
| Flat retainer | Fixed monthly fee | $5K to $25K+ a month | Service may thin as the account grows |
| Bundled CPM (non-disclosed) | One all-in CPM; margin is hidden | Margin often 20 to 50 percent | Agency profits from buying cheaper inventory |
| Hybrid | Retainer plus a reduced percentage | Varies | Generally balanced; check the cap |
| Performance-based | Fee tied to an outcome metric | Varies | Depends entirely on who measures the outcome |
Ranges are typical of proposals we see in the market and vary by scope. Disclosed models, where you see media cost and fees as separate lines, are preferable for any brand that cares where its ads run, because bundled pricing rewards the agency for moving your spend to cheaper supply. For a full cost breakdown, see what a CTV advertising agency costs per month.
What Are the Red Flags When Choosing a CTV Advertising Agency?
- CPMs far below market. Premium streaming inventory with affluent audience data generally clears between $30 and $65 CPM. A proposal at $12 is buying something else.
- Guaranteed ROAS. No one can guarantee return on a channel they have not tested with your brand. A guarantee usually means the measurement is designed to hit the number.
- No named publishers. "Premium inventory" with no list is a description, not a commitment.
- View-through attribution as the only proof. Especially with a 30-day window and no control group.
- A proprietary platform you cannot audit. Proprietary technology can be good. An inability to export log data or verify delivery independently is the problem.
- Reluctance to put fee terms in writing.
- Case studies with no baseline. "312 percent ROAS" means little without knowing how it was measured and what the holdout showed.
Agency, Managed-Service Platform, or In-House?
A CTV agency is one of three ways to run the channel. The right one depends on spend and staffing.
| Option | Best fit | Strengths | Limits |
|---|---|---|---|
| Self-serve platform (e.g., MNTN, Vibe, tvScientific) | Under roughly $25K a month; e-commerce brands | Fast setup, low minimums | Limited supply control; platform measures itself |
| Independent or specialist CTV agency | Roughly $30K to $300K a month | Custom supply, audience, and measurement design; senior attention | Quality varies widely; requires vetting |
| Holding-company agency | $500K+ a month; multi-channel scale | Upfront buying leverage, broad services | Junior staffing on smaller accounts; principal trading is common |
| In-house team | $250K+ a month sustained | Full control and data ownership | Two to four hires, DSP contracts, and 6 to 12 months to mature |
Spend thresholds are approximate. We compare the in-house path in detail in in-house programmatic vs. agency.
How Should High-Consideration Brands Weight the Decision?
For brands with customer lifetime value above $5,000 and sales cycles beyond 30 days, three criteria deserve extra weight.
Environment. A private aviation or wealth management brand is judged partly by where it appears. Content adjacency and publisher quality are part of the message, so supply control outranks CPM efficiency.
Audience precision over reach. When the addressable market is 300,000 households, reaching 5 million is waste. Ask for the estimated size of the target segment and the share the plan will reach, not total impressions.
Patience in measurement. A campaign for a 90-day purchase cannot be judged at day 21. An agency that proposes optimizing to last-click conversions in the first month does not understand the category. Look for a staged plan: qualified reach and site engagement in weeks 1 to 4, inquiry lift in weeks 4 to 10, pipeline and sales after that.
A Scorecard for Comparing CTV Agencies
Score each agency from 1 to 5 on every criterion, multiply by the weight, and total. The weights below reflect what we would prioritize for a high-consideration brand; adjust them to your situation.
| Criterion | Weight | What a 5 looks like |
|---|---|---|
| Supply transparency | 20% | Named publishers, deal types, app-level reports, big-screen minimum |
| Measurement rigor | 20% | Incrementality test designed up front; third-party verification |
| Audience data quality | 15% | Documented sources, definitions, and match rates |
| Fee transparency | 15% | Written waterfall; no undisclosed principal trading |
| Category experience | 10% | References with similar price point and cycle |
| Team and attention | 10% | Named senior team; reasonable account load |
| Contract flexibility | 5% | Short initial term; 30-day exit; data export |
| Reporting and communication | 5% | Agreed cadence; outcome metrics, not just delivery |
A total above 4.0 indicates a strong candidate. Any score of 1 on supply, measurement, or fees should be treated as disqualifying regardless of the total.
How Long Should the Selection Process Take?
Four to six weeks is sufficient for most brands:
- Week 1. Define the objective, monthly budget, target audience, and the outcome metric. Shortlist three to five agencies.
- Week 2. Send the 12 questions and request written responses.
- Week 3. Hold working sessions with the proposed team, and ask each to walk through a sample plan.
- Week 4. Check references and review contract terms.
- Weeks 5 to 6. Select, then agree on the test design before launch.
Agreeing on measurement before the first impression is the single most useful step. It prevents the post-campaign argument about what counts as success. If you have not run the channel before, our guide to whether to test CTV covers the pilot structure.
Frequently Asked Questions
What should I look for in a CTV advertising agency?
Look for documented premium supply, clear audience data sources, incrementality-based measurement, and a written fee breakdown. Those four determine whether your budget reaches the right households in the right content and whether you can prove it worked.
How much should a CTV agency charge?
Most disclosed arrangements run 10 to 20 percent of media spend or a flat retainer, on top of DSP, data, and verification costs. Ask for the full waterfall, because bundled CPM pricing can conceal margins well above those figures.
What is the minimum budget to hire a CTV agency?
Self-serve platforms accept a few thousand dollars a month, while specialist agencies commonly start between $25,000 and $75,000 a month. For a statistically readable test on a high-consideration product, plan on at least $50,000 a month for two to three months.
Put These Questions to Us
Stillwater Media takes on a limited number of luxury and high-consideration brands each quarter, buying premium CTV through private marketplace deals with affluent audience targeting, brand safety controls, and incrementality testing built into every engagement. If your customer lifetime value exceeds $5,000 and you want written answers to all 12 questions, apply for a strategy session.
About Stillwater Media. Stillwater Media is a selective performance media agency for luxury and high-consideration brands, based in Charlotte, NC. We plan and buy premium CTV, programmatic, DOOH, streaming audio, and affluent audience campaigns through private marketplace deals, and we measure them with incrementality testing. Signal. Strategy. Scale.
Frequently Asked Questions
How do I choose a CTV advertising agency?
Evaluate agencies on five things: inventory sources, audience data quality, measurement method, total fees, and data ownership. Ask each for written answers, a publisher-level delivery commitment, and an incrementality test plan before signing.
What questions should I ask a CTV agency before hiring?
Ask what share of impressions will run on named premium services, how frequency is capped across publishers, how incremental sales will be proven, and what the full fee waterfall is. Also ask who owns the DSP seat and data, and who will staff the account day to day.
What are red flags in a CTV agency proposal?
Warning signs include CPMs far below market, guaranteed ROAS, no named publishers, and view-through attribution as the only proof of performance. Refusal to disclose fees or provide app-level reporting is also a reason to walk away.
Is it better to use a CTV agency or a self-serve platform?
Self-serve platforms suit smaller budgets and simple e-commerce goals because they are fast and have low minimums. Brands spending roughly $30,000 or more a month, or selling high-consideration products where environment and measurement matter, usually get more from a specialist agency.
How do CTV agencies prove their campaigns work?
The most reliable method is an incrementality test, such as a geo-holdout that compares sales in exposed markets with matched markets that did not see the ads. Third-party verification and outcome measurement from vendors independent of the agency strengthen the result.
