Amazon DSP vs The Trade Desk: A Luxury Buyer's Comparison

The Amazon DSP vs The Trade Desk decision is usually framed as a question about supply access, and that framing is now largely obsolete. Both platforms reach essentially the same premium streaming inventory — Netflix opened its programmatic supply to Amazon DSP in late 2025, joining The Trade Desk, Google DV360, Microsoft and Yahoo, and most major streamers now sell through every meaningful buying platform. The real difference between them is what each one knows about the household on the other end of the impression, and what it charges you to act on that knowledge.

For a luxury advertiser — private aviation, wealth management, luxury real estate, premium automotive — that distinction matters more than it does for a CPG buyer, because affluent households are a small, high-value slice of a large population and the cost of finding them badly is enormous. Here is how we assess the two platforms at Stillwater Media.

Head-to-head: what actually differs

Fee ranges are what we observe across engagements and negotiated agency agreements; both platforms price on volume and commitment, and neither publishes a rate card.

Where Amazon DSP is genuinely stronger

Purchase-based signal. Amazon's audience data is behavioral in a way no third-party segment is. A "purchased a $4,000+ watch in the last 180 days" audience is built on transactions, not modeled inference from browsing. For luxury goods, premium DTC, high-end consumer electronics and anything with a retail analogue, this is a materially better targeting substrate than the demographic proxies most third-party data vendors sell.

Owned premium video. Prime Video reaches a very large ad-supported base, and Thursday Night Football, live sport and Twitch are inventory pools Amazon controls end to end. When Amazon owns both the supply and the demand path, the intermediary fee layer collapses — which is a real part of why advertisers routinely report lower effective CPMs on Amazon DSP for comparable premium CTV placements.

Amazon Marketing Cloud. AMC, running on AWS Clean Rooms, lets a brand combine its own hashed customer file with Amazon exposure and behavioral signals to run path analysis, overlap studies and audience construction without either party exporting raw records. It is the most capable advertiser-accessible clean room in market for cross-channel measurement, and it is included rather than separately licensed. Our guide to data clean rooms covers the match-rate thresholds that determine whether a small luxury customer file can actually use it.

Momentum. Amazon reported roughly 23% year-over-year advertising revenue growth in Q4 2025 against The Trade Desk's 14%, with The Trade Desk guiding to roughly 10% growth for Q1 2026. Growth rates are not a buying criterion on their own, but they do predict where product investment and supply partnerships go next.

Where The Trade Desk is genuinely stronger

Neutrality. The Trade Desk owns no media. That sounds abstract until you try to run a supply path optimization audit inside a platform that also sells the inventory. When our measurement disagrees with a walled garden's reporting, we need a bidder with no stake in the answer. This is the single most important reason we keep independent DSP capability on luxury accounts.

Identity infrastructure. Unified ID 2.0 and its European counterpart EUID are the most widely adopted alternatives to third-party cookies, with integrations across CRM platforms, CDPs and clean room providers, and a UID2 operator service that can run inside a brand's own AWS environment. For a wealth management or private aviation client whose entire targeting advantage lives in a first-party file, that onboarding path is the asset.

Breadth of third-party data and inventory curation. The platform's data marketplace and curated inventory tooling remain the deepest available for assembling niche affluent audiences from multiple vendors — wealth-screened household data, luxury travel intent, aircraft ownership records, high-value property signals. Our note on wealth-based audience segmentation covers which of those sources hold up under validation.

Measurement flexibility. Independent verification, third-party attribution, custom geo-experiment design and log-level data access are all more straightforward on a neutral platform.

The fee stack, honestly

Neither platform is cheap, and the headline platform fee understates both.

On The Trade Desk, a working impression can carry the buy-side platform fee, a data fee for each audience segment applied, a bid-shading charge on optimized bids, and — on OpenPath supply — a publisher-side fee introduced at 4.5%. Stack those and the difference between gross spend and working media can reach 25% to 35% before the publisher's own take.

On Amazon DSP, the fee stack is simpler on Amazon-owned supply and looks much like any other DSP off it: platform fee, third-party data fees, and supply-side fees on external inventory. Managed service adds a layer, which is part of why the practical minimum sits near $50,000 per month.

One number worth demanding from either platform before you sign: the ratio of gross spend to publisher-received revenue on a representative sample of your actual supply, evidenced by SupplyChain object data rather than platform reporting. Most buyers never ask, and both platforms can answer. On premium CTV bought through clean direct paths we would expect 70% to 80% of gross spend to reach the publisher; anything below 65% indicates an intermediary layer worth removing, and the audit method is the same one we apply in supply path optimization work.

The correct comparison is not fee percentage. It is working media delivered per dollar committed, at a fixed audience quality standard — which requires you to hold the audience definition constant across both platforms and measure what each delivers. That test takes about six weeks and is worth running before any annual commitment.

What we do on luxury accounts

For most clients spending above roughly $75,000 per month on programmatic, we run both. Not as a hedge — as a division of labor:

1. Amazon DSP carries retail-adjacent and purchase-intent audiences, Prime Video and Fire TV supply, and live sport where Amazon holds rights.
2. The Trade Desk carries first-party onboarded audiences, publisher-direct private marketplace deals, wealth-screened third-party segments and reach extension across neutral supply.
3. Frequency and measurement are reconciled outside both platforms, in a clean room or an analytics environment we control, because neither DSP can see the other's delivery.

Below roughly $50,000 per month, running both is a mistake. Split budgets in two learning phases and neither reaches statistical usefulness. Pick the platform that matches the dominant audience signal — purchase behavior or first-party identity — and consolidate.

Onboarding: what the first sixty days actually look like

Platform selection is often argued as though activation were instant. It is not, and the timelines differ enough to affect a launch date.

Amazon DSP. Account setup and creative approval typically run one to two weeks. The longer pole is Amazon Marketing Cloud: instance provisioning, hashed customer file upload, match validation and the first genuinely useful query usually take three to six weeks from kickoff. Purchase-based audiences are available immediately, which means you can be in market fast and measure properly later.

The Trade Desk. Seat access through an agency is quick, but first-party data onboarding is the gating item. A LiveRamp or equivalent identity onboarding cycle — file hygiene, hashing, match validation, segment activation — runs two to five weeks depending on the state of the CRM file. Private marketplace deal negotiation with premium publishers adds another one to three weeks per publisher, and those conversations cannot be rushed by budget alone.

Plan on six to eight weeks from decision to a fully instrumented launch on either platform, and longer if your customer file has never been onboarded before. Brands that compress this usually do so by launching on contextual and platform-native audiences first, then layering owned data once matching completes.

Creative and format capability

The formats a platform supports shape what a luxury brand can actually say.

• Non-skippable premium video is well supported on both, with 15s and 30s standard and 60s available on selected premium supply.
• Interactive and shoppable CTV formats are more mature on Amazon, where remote-triggered actions, QR overlays and add-to-cart integrations tie into the retail environment. For a $200,000 aircraft membership these are largely irrelevant; for premium DTC at $500 to $5,000 AOV they can measurably lift response.
• Pause ads and title-card sponsorships on major streamers have been progressively opened to programmatic partners rather than direct sale only, which brings previously reservation-only luxury formats into DSP-executable inventory.
• Sequential creative — telling a three-part story across a household's exposure path — is executable on both, but requires a stable household identifier, which is more reliable on Amazon's Fire TV device graph and on The Trade Desk where UID2 coverage is strong. Where identity is weak, sequencing degrades into random rotation.
• Dynamic creative optimization is available on both platforms and is, in our experience, overused by luxury brands. Feed-driven variation tends to erode the production values that justify a premium CPM in the first place.

Four mistakes we see in DSP selection

1. Choosing on supply access alone. In 2026 supply is largely commoditized across major platforms; audience signal and fee structure are the real variables.
2. Assuming Amazon's retail data indexes affluence. It indexes purchase. A household buying a $3,000 espresso machine is a strong signal; a household buying household staples at volume is not a wealth signal, and modeled Amazon affluence segments are weaker than the transactional ones.
3. Committing annually before running a controlled test. Both platforms negotiate tiered rates against commitment, and both will let you test first if asked.
4. Ignoring the measurement consequence. A walled garden reporting on its own performance produces flattering numbers. Whatever you choose, keep an independent incrementality read — the design principles are in our comparison of incrementality versus attribution.

The decision, condensed

Choose Amazon DSP if your category has a retail analogue, if purchase behavior is your strongest available signal, if Prime Video and live sport are central to your plan, or if you need a capable clean room without a separate licensing project.

Choose The Trade Desk if your advantage is a first-party customer file, if publisher-direct private marketplace relationships across many streamers matter more than any single owned property, if independent measurement is a governance requirement, or if your affluent audience must be assembled from specialized third-party sources.

Run both if you are spending enough that each platform can reach statistical significance independently, and if you have somewhere neutral to reconcile frequency and outcomes.

If you are evaluating platforms for a brand where customer lifetime value exceeds $5,000 and the sales cycle runs past 30 days, the platform decision should follow the audience strategy rather than lead it. Apply to work with Stillwater Media — we take a limited number of engagements each quarter, and platform selection is the second conversation we have, not the first.

Frequently Asked Questions

Which is better for luxury brands, Amazon DSP or The Trade Desk?

Neither is categorically better; the choice follows your strongest audience signal. Amazon DSP is stronger when purchase behavior is the best available indicator and when Prime Video, Fire TV or live sport are central to the plan, while The Trade Desk is stronger when your advantage is a first-party customer file, when publisher-neutral private marketplace access matters, or when independent measurement is a governance requirement.

What is the minimum spend for Amazon DSP?

Amazon DSP managed service typically requires a commitment in the range of $50,000 per month, while self-service access through an agency seat can start materially lower. Below roughly $50,000 per month in total programmatic spend we would not recommend splitting budget across two demand-side platforms at all, because neither will accumulate enough delivery to produce a statistically usable read.

How much does The Trade Desk cost?

The Trade Desk's buy-side platform fee generally falls between about 12% and 20% of media depending on committed volume, but the headline fee understates the total. Additional layers can include per-segment data fees, a bid-shading charge on optimized bids, and a publisher-side fee on OpenPath supply introduced at 4.5%, which together can push the gap between gross spend and working media into the 25% to 35% range before the publisher's own take.

Can you buy Netflix ads on both Amazon DSP and The Trade Desk?

Yes. Netflix's ad-supported inventory is available programmatically through The Trade Desk, Google DV360, Microsoft, Yahoo DSP and — since the partnership announced in September 2025 — Amazon DSP, with availability rolled out across a dozen major markets including the US, UK, Canada, Japan and Australia. Supply access is therefore no longer a meaningful differentiator between the two platforms.

Should a luxury brand run both DSPs at the same time?

Running both makes sense above roughly $75,000 per month in programmatic spend, and only with a clear division of labor — Amazon DSP for purchase-intent and owned premium video, The Trade Desk for first-party onboarded audiences and publisher-direct deals. It also requires a neutral environment such as a clean room to reconcile frequency and outcomes, because neither platform can see the other's delivery and unmanaged overlap inflates frequency against a small affluent household base.