The Maze: AI shopping is not following one global adoption curve. It is splitting into two operating models. In China, 23% of connected consumers already buy directly through a generative-AI platform. The US reaches 15%; the UK, 12%. The difference is less about chatbot enthusiasm than system design. Chinese platforms can join discovery, product data, checkout, logistics, and service inside one ecosystem. Western AI still behaves more like a clever front door to somebody else's store.
Research use looks global; transaction use does not. Product comparison is already tightly grouped: 35% in China, 30% in the UK, and 29% in the US. That six-point range suggests the habit of asking AI what to buy has crossed markets quickly. Direct purchase produces a different shape. China leads the US by 8 percentage points and the UK by 11. AI can win the research moment almost everywhere while still failing to own the transaction.
China's advantage grows as the task moves closer to commerce infrastructure. Thirty-five percent of Chinese consumers use GenAI to get detailed information on ingredients, features, or product specifications. The US reaches 28%; the UK, 24%. China therefore holds an 11-point lead over the UK on product detail and direct purchase alike. Comparison is the easy layer. Trustworthy catalog data, inventory, payment permission, delivery visibility, and after-sales support are where the model becomes a commerce system.
Alibaba is building one stack; Western players are coordinating several. Alibaba connected Qwen to more than 4 billion Taobao and Tmall listings, plus order management, logistics, and after-sales service. The assistant can move from a natural-language request to comparison, order placement, and delivery management without handing the shopper to a separate environment. That turns conversation into a native marketplace interface rather than a referral channel.
The West is closing the capability gap without removing the handoffs. Amazon combined Rufus and Alexa+ into Alexa for Shopping, bringing comparison, price history, deal finding, cart building, and automated purchases into Amazon's own system. ChatGPT is taking the network route: its commerce protocol connects merchant catalogs from Walmart, Target, Sephora, Home Depot, Wayfair, Shopify, and others. Both models are powerful. But one retailer can control the full stack; an external assistant must negotiate product representation, attribution, permissions, payment, and fulfilment across partners.
The smaller Chinese research-to-purchase gap is the strategic signal. The difference between comparison and direct purchase is 12 points in China, versus 14 in the US and 18 in the UK. These are independent activities, not a measured conversion funnel, so the gaps should not be read as conversion rates. They do show that transactional use sits materially closer to research use in China. For brands, that shifts the optimization target from ranking on a results page to becoming the trusted answer inside a platform that may also execute the order.
Why it matters: Brands need two AI-commerce playbooks. In China, the priority is winning inside vertically integrated ecosystems where the model can influence selection and complete the transaction. In the West, the immediate battle is broader: make product data legible to external assistants, protect retailer relationships, preserve attribution, and stay present when discovery moves outside the storefront. The technology is converging. The channel economics are not.
Sources: Euromonitor | Alibaba Group | Amazon | OpenAI
