Adobe: Amazon and Google Intensify AI Shopping Decision Layer, Sellers Must Adapt to Machine-Readable Rules

During the recently concluded Prime Day event, Amazon placed its integrated AI shopping assistant, Alexa for Shopping, at the core of the deal discovery process. This assistant can push deals, interpret products, track prices, and in some scenarios, directly complete purchases. According to Adobe's statistics, during this event, the conversion rate of traffic directed to retailers through AI assistants was higher than other channels, whereas a year ago, the conversion rate for similar traffic was still below average. This directly reflects the gradual maturation of Amazon's AI shopping tools.

It is reported that AI-driven traffic still accounts for a small proportion of total traffic, with consumers being more cautious in their shopping and highly sensitive to deals. However, the core significance of this change is not that AI dominates the shopping scene, but that it has entered the core position of the product discovery process, and consumers have accepted this model.

Recent announcements from Google at its I/O conference also reveal a similar strategic direction. Its newly launched Universal Cart feature can integrate products and compare deals across merchants, assisting consumers in making choices. The Conversational Attributes feature requires merchants to submit more detailed, hierarchical product information, and the newly added merchant reporting module can track product visibility in the answer engine. Unlike Amazon's closed assistant ecosystem, Google is building an open, cross-scenario layer that can be accessed by other platforms.

While the two platforms adopt different strategies, their strategic direction is consistent. Whether the AI agent operates within a single retailer's closed ecosystem or across platforms based on open standards, the core model remains the same: product information, deal information, and consumer-related data are first integrated and judged by machines, then presented to human users.

This change directly restructures the logic of how products are discovered. AI agents are becoming the first readers of product information, not human consumers. AI agents will judge, based on parsable data, whether a product is understandable, whether it qualifies for a recommendation list, and whether it can be recommended to users. If product data is too sparse, inconsistent, or unreadable by machines, the product will not simply appear lower in search rankings; instead, it will be silently excluded from consideration before consumers even encounter relevant results, a phenomenon known as silent exclusion.

Relevant analysis from Adobe points out that up to 46% of content on some retailers' sites cannot be read by machines, directly limiting product visibility in related scenarios. This is not a search ranking issue; it is an issue of access qualification, and the exclusion process occurs without any notification.

For platform sellers, this change redefines previous operational logic. Over the past two decades, the core task for sellers was to make products findable in search results and to align with human browsing habits for quick page scanning. The new operational requirement now is to make product information interpretable by AI agents that conduct research, filtering, and purchasing on behalf of consumers.

The operational methodology to adapt to this change is called Agentic Commerce Optimization. Its core logic is to treat product data as infrastructure rather than supplementary marketing content. Retailers and brands that can earlier recognize that product data is primarily for machine reading and proactively adjust their product data systems accordingly will gain a competitive advantage.