By Netflix generated $1.5 billion in ad revenue last year, more than double its 2024 haul. And yet the pitch to performance advertisers has always had a hole in it: great content, massive reach, no purchase signal. That changes in Q2 2026, when Amazon Audiences go live on Netflix inventory bought through the Amazon DSP.
Here is what that actually means, what it costs, and whether the numbers hold up.
What just happened
On March 4, Netflix and Amazon Ads announced that U.S. media buyers will be able to apply Amazon Audiences, segments built from shopping, browsing, and streaming behavior across Amazon properties, to Netflix campaigns purchased through Amazon's DSP. The capability launches in Q2 and will expand to additional ad-supported markets later in 2026.
This builds on Netflix's September 2025 decision to open its inventory programmatically in Amazon's DSP across 12 markets. At the time, Netflix was completing what Digiday called its "programmatic integration tour," going live across Google DV360, The Trade Desk, Yahoo, and Microsoft. But Amazon brought something the others could not: commerce data.
Alongside the Amazon Audiences rollout, Netflix is launching a Conversion API (CAPI) for real-time campaign attribution and opening Yahoo DSP deterministic audiences for targeting. Three announcements in one day. Netflix is not tiptoeing into performance anymore.
The CPM reality
I've seen this pitch before. Premium streaming inventory, exclusive audience, massive reach. The deck said $60 CPMs. The invoice said... $60 CPMs. That was Netflix in late 2022.
The good news: prices have come down substantially. According to Digiday, non-targeted in-stream CPMs on Netflix have dropped to the low $20s, sometimes going sub-$20. The average sits around $30, according to a November 2025 Digiday report.
But here is the catch. Ben Vaske, brand media supervisor at Collective Measures, told Digiday that "additional charges for any sort of SSP-applied targeting increase clearing prices to a point where advertisers may need to add up to 40% on top of floor prices to add demo/geo/genre targeting." So that $30 CPM becomes $42 once you layer on the data that makes streaming actually useful for performance.
Now add Amazon Audiences on top of that. Amazon's DSP does offer discounted fees when buying third-party CTV inventory, and its programmatic guaranteed deals carry a 1% take rate that looks generous compared to competitors. In some cases, it could be cheaper to buy Netflix through Amazon than through other DSPs. But "cheaper" and "cheap" are different words. Expect loaded CPMs in the $35-45 range once Amazon's shopping segments are applied.
Is that worth it? That depends entirely on whether the purchase data actually closes the attribution loop. Which brings us to the real question.
What Amazon's data actually gives you
Amazon Ads claims its authenticated identity graph covers 90% of U.S. households. The company says Amazon Audiences draw on "trillions" of data points across shopping, browsing, and streaming behavior. Advertisers will be able to target Netflix viewers based on lifestyle, interests, and products they are actively shopping for.
That is a meaningful upgrade over what Netflix offered before. As Doug Paladino, senior director of CTV and programmatic strategy at PMG, told Digiday: "I wouldn't call what they were running before 'programmatic.' It executed through DSPs, but it didn't bring the promise of programmatic: targeting, measurement capabilities."
The Conversion API is the other half of the equation. In early testing with agency Tinuiti, Netflix's CAPI outperformed benchmarks by more than 75% across financial services, ed tech, and retail categories, according to Netflix and reported by Adweek. That is a single-advertiser test, not an industry-wide result (important caveat), but it is the first time Netflix has had a real performance story to tell.
Harry Browne, VP of TV, audio and display innovation at Tinuiti, told Digiday that clean room testing through Snowflake delivered "really strong cost-per-visit, cost-per-acquisition throughout the funnel" that beat industry performance benchmarks. Again, early days, but the directional signal is positive.
Verdict: the data layer is real, but match rates and actual closed-loop attribution at scale remain unproven. Amazon's 90% household claim is impressive on paper. The question is how much of that graph overlaps with Netflix's 57 million U.S. ad-supported viewers (a figure Netflix attributes to Nielsen). If the overlap is high, you get deterministic purchase attribution on premium CTV. If it is low, you are paying a premium for probabilistic guessing with extra steps.
The scale question
Netflix's ad tier reaches 190 million monthly active viewers globally, a metric the company introduced in November 2025 to account for co-viewing. In the U.S., that is 57 million ad-supported viewers.
For context, Amazon's Prime Video ad tier reaches roughly 130 million people after defaulting subscribers to ads in 2024. Combined through one DSP, that is a lot of living rooms.
The market seems to agree. According to WARC, Netflix's share of the global CTV ad market is projected to rise from 3.7% at the end of 2025 to 9.2% by 2027. Multiple agency executives told Digiday that some advertisers have doubled their Netflix spend since the Netflix Ads Suite launched.
And Amazon's DSP is pulling budget from unexpected places. One anonymous ad exec told Digiday that holdco executives have "moved hundreds of millions of dollars out of The Trade Desk into Amazon."
Verdict: the scale is there. The momentum is real. But Netflix's ad business is still 3.3% of its total revenue. This is a growing business, not a mature one.
What you do not get
Before you rewrite your media plan, know the limits. According to Digiday, program-level targeting (booking spots within specific shows like Squid Game or Bridgerton) is reserved for direct buys. Through DSPs, you get genre, geography, and behavioral targeting, but not show-level placement.
Age-based audience targeting via DSPs is also reportedly not available. And while Netflix has expanded its measurement vendor roster to include iSpot.tv, it may not work with your particular preferred vendor. As Paladino put it: "They have a vendor in each category, but it might not be my vendor in each category."
Frequency management across platforms is still a work in progress. Amazon signed a deal with Omnicom Media in January 2026 to tackle cross-platform frequency capping, but that is one holding company.
Verdict: Netflix through Amazon DSP is not a full-stack replacement for your CTV strategy. It is a powerful addition with real constraints.
What to do Monday morning
If you are a mid-market performance advertiser who has been sitting out CTV because the measurement story was not there, this is your on-ramp. Netflix through Amazon DSP, with purchase-based targeting and the Conversion API, is the closest streaming has come to a closed-loop performance channel.
But run the math first. A $35-45 loaded CPM needs to deliver materially better ROAS than your existing video mix to justify the premium. Request the Tinuiti case study data. Run a two-week test with Amazon Audiences against a control without. And do not let anyone tell you 90% household coverage equals 90% match rate on your specific audience.
The deck will say convergence. The invoice will say premium. Your job is to find out whether the delta between them is worth the reach.
Mira Castellano covers ad tech for The Daily Vibe.
