Streaming TV CPMs: What to Expect in Each Environment

The industry ranges, and what they actually describe

Third-party CTV benchmarking — not Amazon's own published rate card, since Amazon does not publish one — puts the blended connected-TV CPM around $26, with a realistic range of roughly $15 to $45 depending on inventory quality. Premium streaming inventory specifically is reported at $20–$50, premium content adjacency such as inventory sitting next to genuinely appointment viewing at $40–$65, and ad-supported catch-all tiers, comparable to Tubi or Pluto TV-style inventory, at $15–$30. These are industry estimates, not Amazon-specific figures, and they should be treated as planning ranges rather than a quote you can hold anyone to.

It's worth being precise about where these numbers come from, because it changes how much weight to put on them. They're aggregated across the whole connected-TV advertising market — every DSP, every publisher, every buying method — not specific to Amazon DSP's auction dynamics. Amazon's own inventory could plausibly sit anywhere inside or even outside that range depending on demand for a specific supply source at a specific time, which is exactly why a planning range is more useful here than a single number.

Seasonality shifts these ranges too, independent of anything about the inventory itself — CPMs across the whole category tend to rise heading into the fourth quarter as more advertisers compete for the same holiday-season attention, which is worth budgeting for separately from any account-specific trend a brand might otherwise misread as a performance change.

Why a blended DSP CPM looks nothing like a streaming CPM

This is the single most common source of confusion in budgeting a streaming test. Across 27 advertisers in our own Amazon DSP book over 31 days this summer, connected-TV and streaming inventory specifically blended to a $15.51 CPM — toward the lower end of the reported premium range. But the account's overall DSP CPM across every supply source, including much cheaper inventory, was far lower still: Alexa device inventory in that same period ran a $0.32 CPM. A brand that sees a low blended DSP average and assumes streaming specifically is cheap is about to be badly surprised the first time they try to buy streaming volume at that rate.

The same gap shows up going the other direction. A brand that only reads its account's premium content-adjacency CPM — the number closest to the top of the reported range — and assumes that figure describes the whole account will systematically over-budget for the cheaper inventory sitting alongside it. Either mistake comes from reading one number and assuming it describes the whole account, when a modern Amazon DSP book, touching well over a hundred distinct supply sources in a typical month, simply doesn't have one CPM.

A worked example

Say a brand wants to know what $30,000 buys on Prime Video and Fire TV Channels combined. At a representative $25 CPM blended across the two (Fire TV Channels' broader run-of-service inventory pulling the average down from Prime Video's premium-content rate alone), that's 1.2 million impressions. Compare that to what the same $30,000 would buy against premium content adjacency at $50 CPM — 600,000 impressions, half the volume, for inventory sitting next to genuinely appointment content rather than broader run-of-service programming. Neither number is right or wrong; they're different trades between reach and content premium, and the correct choice depends on the campaign's actual objective.

What to do when a quoted CPM looks wrong

If a CPM you're seeing in-platform is dramatically below the industry ranges above, check what supply sources are actually included in that line item — a blended campaign spanning cheap owned-and-operated inventory alongside premium streaming will report an average CPM that misrepresents both pieces. If a quoted CPM is dramatically above the ranges, check whether it's genuinely premium content adjacency, like live sports, before assuming you're being overcharged; that band is real and reported consistently across the industry.

If a CPM shifts sharply week to week with no change to your own targeting or budget, check the calendar before assuming something broke. Marquee live events and major retail weeks concentrate demand across the whole platform at once — the same dynamic that has been separately measured to roughly double DSP click costs across our own book during Amazon's own Prime Week — and a CPM spike during a comparable high-demand week elsewhere on the platform is closer to expected than to an error worth escalating.

The common mistake

The mistake is budgeting a streaming test off a blended account CPM from a prior campaign that included cheap inventory the new campaign won't touch. A brand that saw a $4 blended DSP CPM last quarter and budgets $10,000 expecting 2.5 million streaming impressions this quarter will be off by an order of magnitude. reMKTR budgets every streaming line item off supply-source-specific CPM data, not a blended account average, specifically because the gap between the two is large enough to break a media plan built on the wrong number.

A second, less obvious mistake is treating a low CPM as automatically the better outcome. A cheap CPM against the wrong audience is money spent efficiently on the wrong thing — the actual question is cost per outcome that matters, whether that's completion rate, branded search lift, or new-to-brand rate, not cost per raw impression on its own.