AI dominates China's "Double 11" festival, but shoppers still wait for a tru...

by CHENG Lu

China's Double 11 shopping festival — the world's biggest online retail event — has become a testing ground for how artificial intelligence is reshaping e-commerce.

As this year's two-week sales marathon ends, frustration has flooded Chinese social media over "price traps": products that cost more after pre-sale deposits or livestream discounts. Even as Tmall, Alibaba's flagship marketplace, issued about 50 billion yuan (about US$7 billion) in coupons, shoppers complained that "Double 11 isn't cheaper anymore."

Behind the confusion lies algorithmic targeting. Alibaba's Smart Savings Engine, powered by generative AI, tailors coupons to each user's behavior. If a shopper lingers on a product or repeatedly revisits a store, the system automatically sends discounts. During trials, conversion rates rose 15 percent — showing that AI now determines not only what consumers see but how much they pay.

Platforms get smarter — consumers, more confused

This year marks China's first large-scale commercial rollout of generative AI in retail. Alibaba, JD.com, ByteDance and Tencent have embedded AI into search, advertising, logistics and customer service, making algorithmic precision the new measure of competitiveness.

ZHANG Kaifu, head of search and promotion at Alibaba China Commerce Group, said Taobao's core systems — search, recommendation and ads — have been rebuilt around large-language models. "Every key metric has posted double-digit improvement," he told Jiemian News.

Most upgrades remain invisible to users. New AI shopping guides and chat functions are layered onto existing Taobao features rather than replacing them.

WANG Yunfeng, CTO of Zhi-Tech Group, said platforms are trying to balance innovation with control. "They want AI to enhance user experience and efficiency, but must also protect data and business models," he said. Tools such as the Model Context Protocol (MCP) allow AI to tap external data sources, though adoption remains limited.

For Alibaba, even small efficiency gains matter. Advertising and sponsored search generate much of its customer-management revenue, so higher click-through and conversion rates translate directly into profit. Zhang compared Taobao's AI strategy to Google's — embedding intelligence across every user touchpoint rather than building a single chatbot.

The race for the next entry point

AI is also becoming a new entry point for consumer decisions. "AI is evolving into the next interface for shopping," Wang said. "Beyond apps like Deepseek, Kimi, Yuanbao and Quark, future access points could be phones, wearables or home devices."

In China, early "deep-research" tools can already generate personalized design plans, suggest products and compile purchase lists — a preview of what a true AI shopping assistant could do.

Outside China, OpenAI's ChatGPT has gone further. Integrated with Walmart, Etsy and Shopify, it lets users search, compare and pay directly in chat. ChatGPT has about 700 million weekly active users and handles around 75 million shopping-related conversations weekly, forming what analysts call a "conversational commerce" loop connecting recommendation, decision and payment.

ByteDance's Doubao and Tencent's Yuanbao are building similar ecosystems within Douyin E-commerce and WeChat Shops. Research firm Gartner says China's "AI super apps" are taking shape, with WeChat and Douyin likely to embed AI deeper across their platforms.

Within Alibaba, insiders say user behavior is already shifting as tools like Taobao AI Universal Search and Quark AI roll out. "consumer behavior will not look the same as before," said a person familiar with the company's plans.

Openness versus control

The future of AI commerce may depend on whether companies can break out of their data silos. A universal shopping assistant would need access to inventories, prices, reviews and payment systems across platforms — data each company treats as its competitive moat.

In the United States, OpenAI's Agentic Commerce Protocol (ACP) lets small retailers plug directly into ChatGPT. The model can already process PayPal payments inside conversations, completing the "chat-to-checkout" loop. Each integration adds pressure on traditional e-commerce gateways.

Chinese firms remain cautious. Their AI deployments still focus on internal efficiency rather than re-engineering the retail ecosystem. But analysts expect growing interoperability as competition intensifies, and regulators push for common data standards.

For now, AI in China's e-commerce is about precision, not simplicity. The all-in-one digital assistant that can navigate the nation's biggest shopping festival remains a vision for the future— but momentum is building. As AI begins to weave data, payments and logistics into a single fabric, the lines between platforms are starting to fade.