How ChatGPT ranks products, and why AI agents have retailers rethinking search
ChatGPT's built-in shopping features are already influencing product discovery, and the ranking logic is unlike Amazon's or Google's. Products are selected based on what the model infers the user really wants, drawing on the query, saved preferences and custom instructions. Details like pricing, descriptions and reviews are sourced from third-party providers rather than straight from brands, the model decides which attributes matter most for each request, and results must pass internal safety checks. There are no paid placements: results are generated independently, displayed in a visual carousel with image, price and simplified description, sometimes with unverified labels like 'budget-friendly' and AI-generated review summaries. Merchants cannot upload or optimize listings directly, though they can submit to be indexed. The takeaway for sellers: structure product data cleanly across marketplaces, build presence on the third-party platforms AI pulls from, and position products around shopper intent.
Amazon's compliance AI is now watching beyond its own platform. One seller received a compliance warning over a non-compliant health claim found not on their Amazon listing but on their own direct-to-consumer website, flagged automatically. Amazon's bots are scanning brand websites, other marketplaces, product images via text recognition, marketing claims and pricing inconsistencies, and off-Amazon content can trigger on-Amazon penalties including delistings. Sellers are advised to audit websites and imagery with the same rigor as their listings.
Walmart is preparing for a future in which AI shopping agents act for consumers. The retailer is building its own in-app agents for tasks like reordering groceries, while bracing for third-party agents such as OpenAI's Operator that choose products, apply preferences and complete checkout. These bots ignore beautiful imagery and emotional hooks, responding to data, accuracy, structure and price, which threatens traditional SEO and advertising tactics and could erode retailers' direct customer relationships. Walmart expects agent-to-agent communication protocols to emerge, and advises structuring listings for clarity, keeping prices competitive and anticipating the bot as a new kind of buyer.
A widely shared analysis of SEO in May 2025 paints a mixed picture. Headwinds: Google's AI Overviews are cutting organic click-through rates by 20 to 35 percent depending on the study, 80 percent of top results are controlled by large publishing groups, independent content sites are fading, SEO job listings are down 37 percent year over year, and link building is getting harder. Positives for e-commerce brands: search still drives 43.8 percent of site traffic against 0.1 percent from AI bots, Google processes over 5 trillion searches a year and growing, Google Discover is expanding to desktop, and disciplined SEO still produces winners. The practical guidance: diversify traffic beyond organic search into email, social, communities and AI agents; invest in experience, expertise and trust signals; structure content to surface inside AI Overviews; and compete on forums and user-generated platforms that still rank.
Logistics check: the Drewry World Container Index rose 8 percent to $2,233 per 40-foot container, still 78 percent below the September 2021 pandemic peak but 57 percent above the 2019 pre-pandemic average. Shanghai to New York rates jumped 19 percent to $4,350 and Shanghai to Los Angeles 16 percent to $3,136. Quick hits: Amazon put new limits on Seller Fulfilled Prime eligibility, revealed several new fulfillment robots, and Walmart's e-commerce sales jumped 21 percent. Trivia: back-in-stock emails are the best-converting e-commerce email type at 5.84 percent, well ahead of cart abandonment emails at 2.56 percent.
