As competitors dial back advertiser control in pursuit of full automation, Amazon Ads is positioning its new AI tools as transparent alternatives.
In May this year, Mark Zuckerberg said that he wants advertisers to hand over their bank accounts and walk away. In his vision for Meta’s advertising future, brands would provide "no creative, no targeting, no measurement"—just an objective and payment credentials. The algorithm handles everything else.
Google has pursued a similar path since 2021 with Performance Max, consolidating inventory across Search, Shopping, YouTube, and Display into algorithm-managed campaigns. While over one million advertisers now use the PMax product, it has vehement detractors, particularly the more sophisticated advertisers who want to closely manage performance. Google rolled out a series of updates this year that appeared to be an about-face, granting advertisers more control again.
At Amazon’s Unboxed event, the company staked out opposing territory in its raft of product announcements. In announcing a unified campaign manager, AI-powered Ads Agent, and Creative Agent tools, Amazon is explicitly positioning itself against full automation – what Kelly MacLean, the vice-president leading Amazon’s ad-tech offerings, calls operating “in more of a crystal box, not a black box."
Amazon is collapsing its demand-side platform (DSP) and Sponsored Ads products into a unified campaign manager. The company says this enables real-time signal sharing across the customer journey that was previously impossible when the systems operated separately.
“When Sponsored Ads campaigns share signals in real-time with Amazon DSP campaigns, the system enables smarter cross-funnel optimization that activates at the exact moment of customer engagement,” MacLean said. The platform processes what Amazon describes as “trillions of shopping, browsing, and streaming interactions” to coordinate campaigns across its owned and partner inventory.
Advertisement
Setup time drops by over 60%, from more than an hour to under 10 minutes, according to Amazon. Advertisers upload a media brief; the system translates it across display, video, audio, and sponsored ad formats.
That’s automation, and it indeed could be a boon to less sophisticated advertisers as well as larger advertisers with extensive product assortments. But the transparency claim rests on what happens after setup.
Advertisers using Performance+ and Brand+—Amazon’s AI-driven campaign products launched earlier this year—retain what the company calls “optimization levers” and can see where ads delivered on the backend. In a May white paper, Amazon described this as a “glass box” approach (MacLean uses “crystal box”), outlining how its predictive AI model transforms behavioral signals into targeting decisions using “customer and ad embeddings.” For sophisticated advertisers accustomed to manual campaign management, that may provide sufficient visibility.
Another significant feature announced today is Ads Agent, which Amazon describes as the first AI-powered tool for campaign orchestration across planning, execution, and measurement. It operates through natural language rather than dashboard navigation.
Advertisers can query: “Show me all campaigns within a specific budget with delivery rates below 75%.” Or input a targeting brief and have the system draft the strategy using Amazon’s audience data. The advertiser reviews and adjusts.
A notable new capability: natural language access to Amazon Marketing Cloud (AMC) analytics. Previously, extracting insights from AMC required data scientists who could write SQL queries. Ads Agent translates conversational questions into SQL automatically.
MacLean positions this as “an AI partner and co-pilot—enhancement rather than replacement.“ However, the practical reality is that most advertisers lack the expertise to determine whether the SQL query generated by the AI accurately reflects their strategic intent. Compare that to Meta’s vision, where even the creative disappears into the black box.
Transparency matters more in retail media than in social or search advertising. When the same company controls both the advertising platform and the marketplace where advertised products compete for visibility, brands question whether the algorithm optimizes for advertiser performance or platform revenue. Amazon faces this skepticism more than most, and transparency becomes essential to maintaining advertiser trust. Amazon’s response is technical: insight cards and machine learning recommendations that help marketers “audit their campaign performance.” Whether that’s sufficient depends on what advertisers are actually able to verify.
Amazon’s new Sponsored Product Video ads reveal the duality in what Amazon is proposing: transparency in reporting and personalized content. Sponsored product Video Ads - a new ad type cut from the cloth of the company’s highly efficacious Sponsored Brand Video ads, appear in search results and on product pages, with no maximum length. The innovation: the actual highlighted content of the video is dynamically tailored by audience.
Advertisement
Three shoppers viewing the same product might see three different videos emphasizing different features based on predicted preferences. That could be very effective personalization. But it also means advertisers may not see the creative their customers see.
How does an advertiser audit creative performance when the creative itself is variable and audience-specific? Amazon provides campaign-level reporting. But when AI is generating audience-specific creative variants at scale, “transparency” becomes a question of granularity. You can see aggregate performance; you can’t necessarily see which specific creative resonated with which audience segment and why.
Amazon plans to launch Full Funnel Campaigns in 2026—AI-driven setups that coordinate awareness, consideration, and conversion in unified campaigns that take “just a few clicks” to create. The system recommends personalized approaches based on each brand’s opportunities.
This is where the crystal box faces its most significant challenge. When AI simultaneously optimizes across awareness spend, consideration tactics, and conversion media—balancing reach, engagement, and sales across owned inventory and third-party publishers—transparency becomes exponentially more complex.
Advertisement
Amazon’s May white paper describes how its predictive AI uses "customer and ad embeddings" to forecast intent and match ads to likely converters. That’s a detailed methodology for data scientists. For most marketers, it might as well be a black box with technical documentation.
The distinction between a glass box and a black box may ultimately be whether you're shown the algorithm's decisions and whether you can actually evaluate those decisions meaningfully.
<!-- inArticleBlock -->
Amazon appears to have studied the backlash against Performance Max and calculated a differentiated position. Rather than rushing to match Meta's zero-input automation, the company is claiming middle ground: automation for efficiency, transparency for trust.
What’s clear so far is that Amazon Ads is transforming from a suite of tools into a self-optimizing ecosystem. And they are attempting to answer to advertisers’ biggest pain point: scale and sophistication. The tricky part is delivering all of that without opaque automation. For now, Amazon is betting that advertisers exhausted by competitors’ opacity will choose a platform that at least shows them the algorithm’s work—even if they can’t fully evaluate it.
Amazon Ads
Amazon Ads helps brands design ad experiences that delight customers and deliver meaningful results. With our first-party insights and hundreds of millions of global...
Amazon