Amazon generated US$19.64 billion in advertising services revenue in the third quarter of 2026, putting its media business close to the US$20 billion quarterly mark. In its October 29 earnings disclosure, the company also pointed to AI-driven advertising tools as part of its plans to make buying across formats easier.
The two developments deserve to be examined separately. Advertising revenue shows what the platform is earning today. Amazon Ads Agent, the buying system unveiled in September, shows how Amazon wants advertisers to plan and optimize future spending. Neither figure on its own establishes whether an AI-managed campaign produces a better outcome for a particular brand.
Table of contents
Jump to each section:
- Q3 revenue growth comes with a calendar caveat
- Amazon Ads Agent changes who executes media buys
- Claims of better performance need an independent test
- The transparency question extends beyond Amazon
- What this means for marketers
Q3 revenue growth comes with a calendar caveat
Amazon’s third-quarter advertising business continued to grow from the previous year, even as growth slowed relative to its second quarter. The company reported advertising services revenue of US$19.637 billion, compared with US$17.703 billion a year earlier. It said growth excluding currency changes was 12%.
US$19.637 billion in Q3 2026 advertising services revenue, approximately 11% higher on a reported basis year over year. Amazon separately reported 12% growth excluding foreign-exchange effects, according to its quarterly earnings exhibit.
The sequential comparison looks less impressive: Amazon Ads generated US$19.809 billion in the second quarter. But treating that small quarter-to-quarter decline as a demand signal would miss a major calendar change. Amazon moved its Prime Day sales event into the second quarter in 2026, whereas the event fell in the third quarter a year earlier.
That distinction matters for media buyers. Retail media revenue is affected by the promotional calendar, inventory and consumer shopping patterns. A platform’s financial trend can indicate where spending is accumulating, but it does not reveal whether an individual advertiser’s acquisition cost, sales incrementality or return on spend improved.
Amazon expects advertising revenue to grow at a high-teens year-over-year rate in the fourth quarter, excluding exchange-rate effects. That is its forecast, not a realized result, and should not be treated as a measure of advertiser ROI.
Amazon Ads Agent changes who executes media buys
Amazon is positioning its previously announced Ads Agent as a way to make media buying less dependent on navigating separate campaign systems. The agent sits across sponsored advertising, display, video and audio, with buyers able to ask questions, request recommendations and take some actions using natural language.
The important change is not that a chatbot can describe a dashboard. It is that the platform can increasingly choose how a campaign is assembled and optimized after advertisers set commercial objectives. Full-Funnel Campaigns combine sponsored ads with upper-funnel inventory, while DVA+ brings display, video and audio planning into a more unified buying experience.
The familiar promise of automation is less time spent on routine setup. The strategic trade-off is a larger share of decisions taking place inside the same company that sells the advertising. Simpler execution can be genuinely valuable, but it also changes where marketing teams need visibility and control.
Amazon says advertisers retain choices around budgets, creative approval and objectives. DVA+ also offers advanced settings for more detailed control. Its simplified full-funnel option, however, leaves more of the channel mix and audience decisions to automated systems. These configurations should not be described as having identical oversight.
Claims of better performance need an independent test
Amazon has published promising early results for Full-Funnel Campaigns, but those claims predate the latest quarter and cannot be used to explain its reported ad revenue. They are product-test claims, not a public measurement of Amazon Ads Agent’s contribution to company growth.
67% higher long-term return on ad spend and 29% lower new-to-brand acquisition cost. Those are Amazon’s reported comparisons for Full-Funnel Campaign beta users against manual campaign setups, according to Amazon Ads’ September product announcement. The results are company-reported, not independently verified outcomes for all advertisers.
The distinction is especially important because the platform’s long-term sales measurement can incorporate forecasts based on shopping behavior alongside actual sales recorded over time. Such estimates may be useful for planning, but marketers still need to understand what counts as a verified purchase, what is modeled, and how a reported lift was established.
Reported revenue growth proves the media platform is selling more advertising than a year ago. It does not prove that automation is adding incremental sales for customers. A campaign can look more efficient inside one reporting system while shifting sales that would have occurred anyway, or while allocating more budget toward outcomes measured primarily by that same platform.
The transparency question extends beyond Amazon
In comments on automated media buying, Amazon Ads executive Amit Bhattacharyya acknowledged the need for more disclosure: “We are figuring out how to give them more transparency on the audience side.” His remark captures an unresolved question for AI-led campaign systems: buyers can approve the objective while having limited insight into how each allocation choice was made.
The trade-off extends across platforms, although the products are not identical. Google’s current AI search advertising guidance describes placements that automatically draw on existing Search, Shopping and Performance Max campaigns, with performance included in aggregate reporting rather than a standalone AI Mode view.
| Dimension | Amazon Ads Agent | Google AI Mode ads |
|---|---|---|
| Primary change | Automated planning and optimization across Amazon ad formats | Existing Search and Shopping campaigns become eligible for conversational placements |
| Advertiser control | Objectives and approved creative, with varying advanced campaign controls | Existing Google Ads controls apply, without a separate AI Mode opt-in |
| Measurement consideration | Understand modeled long-term sales and inventory allocation | AI Mode outcomes are included in aggregated campaign reporting |
These systems solve different problems. Amazon is reshaping how advertising is bought across its media portfolio; Google is expanding where a conventional campaign can appear. But both make parts of the buying or placement decision less directly visible to the marketer.
The next question for automated advertising is not simply how much manual work disappears. It is whether buyers can verify what the automation decided, what it cost, where ads appeared and whether the result was incremental.
What this means for marketers
Amazon’s latest earnings show a substantial advertising business investing in a different buying model. Marketing teams assessing that model should separate four questions that are often bundled together as a single claim of AI efficiency.
Read the financial comparison carefully. Prime Day’s changed timing complicates the latest sequential comparison. Amazon’s advertising revenue should not be treated as a direct proxy for advertiser return.
Define where human approval matters. Determine which platform decisions a buyer can review before launch, which the system may automate later, and whether the choice differs by campaign type.
Separate modeled from observed outcomes. Ask how long-term sales, new-to-brand customers and incremental lift are calculated, especially when performance claims come from the platform selling the media.
Preserve a test outside the platform. Where possible, use first-party sales data, holdout groups or independent measurement to check whether the recommended spend creates value beyond what existing campaigns would have delivered.
The traditional assumption is that automating a buying system primarily removes friction from work the marketer already understands. The reality may be more consequential: it can move budget allocation and optimization judgment from a human team to the supplier of the inventory.
That is not automatically a reason to reject agentic advertising. It is a reason to define the terms of oversight before handing over more decisions. Buyers may accept less manual control when they can establish that business outcomes are better and the process remains auditable.
As advertising becomes more automated, accountability may become a stronger differentiator than ease of use. A platform that simplifies buying and allows marketers to test its claims independently can earn a different kind of trust from one that only promises efficiency.
