Amazon Prime Day drives 15–40% ad spend surge, DSP and CTV gains

Full analysis

Amazon Ads is pulling brand budgets up the funnel. Media buyers at Wpromote, Power Digital, Tinuiti, and PMG report client spend on Amazon up 15 to 40% year-on-year ahead of Prime Big Deal Days (Oct. 6 and 7), with full-year 2026 growth near a third. The incremental money is landing in Amazon's DSP and in connected-TV inventory like Prime Video and Fire TV, not in the sponsored-product search ads that built the business. Amazon's Q2 2025 ad revenue hit $19.8 billion, up 26%.

This is easy to undo on any single campaign and very hard to undo as a category. A brand can pull a Prime Video test next quarter. A holding company that formalizes Amazon upfront commitments and builds its measurement stack around Amazon Marketing Cloud cannot. What's actually being decided is whether the open-web DSPs and SSPs keep a seat at the brand-budget table, or whether Amazon becomes the default full-funnel buy for non-endemic advertisers. The deadline is the 2027 upfront planning cycle, when these commitments get locked.

The Market Analyst

Amazon Ads at $19.8 billion a quarter growing 26% is bigger than most of the open programmatic web put together. That scale is the story. The share is coming out of Google search and Meta social, and the holdco trading desks at Omnicom, WPP, and Publicis are racing to formalize Amazon upfront commitments to grab volume discounts. Those discounts buy dependency. For The Trade Desk, this directly dents the retail-media pitch, because Amazon now sells the same full-funnel bundle with a purchase signal no independent can match. Magnite and PubMatic feel it on CTV yield. For a general reader: the one retailer that knows what you actually bought is now selling the TV ad and grading its own homework.

The Skeptic

A 15 to 40% range says almost nothing. That spread is the gap between a client who doubled down and one who nudged a line item. The real question is whether this Amazon brand money is new or just Google and Meta budget wearing better attribution. Amazon's incrementality tools are self-reported and self-validated. Independent marketing-mix studies keep finding Amazon's halo effects overstated at the category level. And the FTC is investigating Amazon's ad auction. If that auction is found to self-preference, the clean-room measurement story that justifies all this spend gets a lot harder to defend. "Defensible to the CFO" often means nobody has stress-tested it yet.

The Operator

Buyers are now running two mandates on one platform: defend lower-funnel return on ad spend while spending up-funnel on DSP and CTV. That only works if the incrementality measurement is built in before launch. The agency that can show a Prime Video dollar lifted sponsored-product conversions through an Amazon Marketing Cloud lift study keeps the budget. The one retrofitting it loses the conversation. At 90 days, the squeeze lands on trading desks without a direct Amazon DSP seat arrangement. They get priced out while the independents who locked in preferred terms keep the volume. For the non-specialist: whoever can prove the TV ad sold more product gets next quarter's money.

The CFO

The spend story is simpler than it looks. Spend is moving to Amazon because it is easier to justify, not because it provably works better. A closed-loop purchase signal is the best-looking number in the room, and finance rewards the number it can see. That is also the trap. When one platform owns the inventory, the measurement, and the retail data, it sets the take rate and you have no outside yardstick to argue it down. The margin compression hides inside Amazon's own fees, which do not show up on the surge chart. The payback looks clean right up until there is no competitor to benchmark against.

Tensions

Three real disagreements. First, is this spend incremental or cannibalistic? The Market Analyst says Amazon is taking share from Google and Meta, which makes it a genuine reweighting. The Skeptic says it may be the same brand dollars chasing a prettier dashboard, which would mean nothing structural changed. Second, is the measurement advantage durable or borrowed? The Strategist's closed-loop moat and the CFO's "best-looking number" are the same fact read two ways, one as a lasting edge and one as a dependency nobody has priced. Third, the FTC. Everyone treats the auction investigation as background noise. If it forces auction transparency, the self-validated measurement story is the first thing to crack.

Synthesis

This hinges on two beliefs. One: that Amazon's incrementality measurement is both real and will stay proprietary. Two: that the up-funnel spend is new money, not reshuffled Google and Meta brand budget. The council leans toward the shift being real at the scale level and oversold at the proof level. Amazon is genuinely winning brand budget, but "defensible to the CFO" is doing the persuading, and no independent measurement confirms the lift the buyers are selling internally. Before committing upfront dollars, run an independent marketing-mix read against the Amazon Marketing Cloud lift study. If the two disagree, you are buying a dashboard, not a result. The thing to de-risk is the single-source grading: the whole case rests on trusting the seller's own numbers during the exact window when the FTC is asking whether those numbers are rigged.

Prediction: Amazon's full-year 2026 ad revenue, reported in its Q4 2026 earnings on the late-January 2027 call, will grow faster than 20% year-on-year, pulling further ahead of both The Trade Desk and Magnite on revenue growth.

Confidence: Medium. Q2 2025 already ran 26% and brand budgets are shifting up-funnel now.

Why: Amazon posted $19.8 billion in Q2 2025 ad revenue at 26% growth, and buyers across Wpromote, Tinuiti, PMG, and Power Digital report 2026 client spend rising 15 to 40%, with the incremental money moving into higher-priced DSP and CTV inventory rather than cheap sponsored-product clicks. Moving budget up-funnel raises revenue per impression, so the growth rate holds even if unit volume flattens. The opposite outcome, growth slipping below 20%, would need either the FTC investigation to force a margin-cutting auction change before year-end or a macro pullback the buyers explicitly say they are not seeing, and consent decrees do not move that fast.

Revisit by 2027-02-15: We're right if Amazon's reported 2026 ad revenue grows more than 20% year-on-year and tops the revenue growth rates The Trade Desk and Magnite report for the same year. We're wrong if Amazon's growth comes in at 20% or lower, or if either The Trade Desk or Magnite posts a faster 2026 growth rate.

The benchmark names are the two public pure-plays whose retail-media and CTV pitch competes most directly for the same brand budget, so if Amazon out-grows both while taking that spend, the share shift is real and not a measurement story.