研究显示ChatGPT超八成购物决策源自产品卡片

Overall: AI shopping research shows that product cards play a decisive role in purchase decisions; consumers should understand how they work and avoid blindly following the first recommendation.

1. Over 80% of purchase decisions originate from product cards: In ChatGPT, 84% of final product selections came from cards embedded in the conversation; products not shown as cards were almost never chosen.

2. Position has a huge impact: The first card is selected 43.4% of the time, higher than the 29% random baseline, while the second card is selected below random; 76% of deal selections go to the first position. Consumers should browse multiple cards rather than fixating on the first one.

3. Card information directly shapes choices: Images, price, stock, and delivery determine the next step, and users have also found price mismatch cases; before ordering, compare with the product detail page.

4. Practical tip: During AI shopping, actively review multiple cards and verify prices and specifications so your decision is not swayed by display order.

Overall: In AI shopping scenarios, product cards have become the critical entry point for brands to win purchase decisions; position and consistency are the keys to marketing.

1. User behavior observations: 84% of final choices in ChatGPT come from product cards, and 75% of tasks begin with card exposure; images, titles, and prices influence decisions early.

2. Implications for brand marketing: Getting exposure is only the starting point; the real conversion battle happens at the product-card and deal-card stage. Brands need to push their products into the card display pool and compete for the top position.

3. Pricing and price competition: 76% of deal selections land on the first deal card, and price, stock, and delivery information directly determine users' next steps. Brands must ensure prices and specifications are consistent across product cards, deal cards, and product detail pages to avoid mismatches.

4. Channel-building reference: The value of product cards is similar to Amazon's Buy Box. Brands can treat product cards as a core asset and adapt to AI channels like ChatGPT through agentic commerce optimization (with 8 practical recommendations).

Overall: The rise of AI shopping creates new traffic entry points and growth markets. Sellers should master product-card optimization to seize first-mover opportunities.

1. Growing market: The study covers ChatGPT and Google AI Mode, with 40 users completing 224 shopping tasks across 6 product categories, showing that AI is becoming a real shopping-decision environment worth early investment.

2. Responding to the shift and opportunities: The research introduces the concept of agentic commerce optimization and provides 8 practical recommendations, with the core being to get products into the product-card display pool and compete for the first-card position.

3. Positive effects and risk warning: The first card is selected 43.4% of the time, and the first deal slot captures 76% of deal selections, showing highly concentrated positional value. Products that fail to enter the display pool will miss most comparison opportunities, a clear risk.

4. Takeaways: Sellers can study the report's session transcripts, user decision references, and price-mismatch cases to learn how users compare and exclude options, and then adjust the appeal of their product cards.

Overall: AI shopping depends on product cards to present product information and images, which raises new requirements for product design, production, and digitalization.

1. Product production and design requirements: Product cards need clear images, selling prices, and core specifications; factories should consider information standardization during design and production so products fit card presentation.

2. Business opportunity: Front-of-card information has a real influence early in the decision process. Products with complete specifications and clear parameters are more likely to be quickly selected by AI assistants and users.

3. Digitalization and e-commerce implications: The study uses full behavioral tracking and think-aloud reconstruction, suggesting that factories should build structured product data to prepare for AI recommendation pools and product-card display pools.

4. Alignment points: Sellers need to ensure products accurately match the corresponding product cards and aim for higher ranking in the deal-card list; factories can achieve this in coordination with brands and sellers.

Overall: The AI shopping study provides quantitative evidence of the value of product-card entry points, offering service providers new business directions and methodologies.

1. Industry trend: AI shopping behavior is becoming a new arena, and product-card value can be benchmarked against Amazon's Buy Box; service providers can build offerings around product cards and deal cards.

2. New research and methods: The study uses full behavioral tracking plus think-aloud reconstruction across multiple categories and tasks, a model that can be used to evaluate user decisions in AI shopping.

3. Client pain points: Brands and retailers commonly face clear challenges: difficulty entering the product-card display pool, difficulty getting the top position, and inconsistent card information.

4. Solutions: ReFiBuy defines the relevant optimization as agentic commerce optimization and provides 8 practical recommendations. Service providers can use these to deliver implementation services covering exposure, ranking, and consistency maintenance.

Overall: In AI shopping, product cards take on a Buy Box-like decision function; marketplace operators should pay attention to positional factors and card information management.

1. Business demands on platforms: Product cards directly display images, prices, and specifications, and are the source of 84% of final selections; sellers need platforms to provide fair and predictable card display and ranking mechanisms.

2. Operational management: The first card is selected 43.4% of the time, while the second card is below random; positional advantage is heavily skewed toward the first slot. Platforms should account for this bias when adjusting rankings, balancing efficiency and fairness.

3. User behavior observations: Price, stock, and delivery information directly determine users' next steps, and 76% of deal selections go to the first slot. Platforms can optimize card information structures to improve decision experience.

4. Risk mitigation: The full report includes users' responses to sponsored display slots and price-mismatch cases. Platforms can use these to set advertising slot rules and information-consistency checks to avoid losing user trust.

Overall: The AI shopping behavior study offers quantitative evidence of the value of product-card entry points, revealing new industry trends and open questions worth exploring.

1. New industry trend: Product cards in AI chat interfaces have become the core entry point for shopping decisions. By comparing product-card value to Amazon's Buy Box, the study suggests that AI shopping may reshape traffic allocation logic.

2. Research method innovation: The study does not rely on surveys or traffic statistics; instead, it uses full behavioral tracking and think-aloud reconstruction, covering 40 participants, 6 categories, and 224 tasks, providing a new paradigm for AI interaction research.

3. New questions: Positional factors create substantial decision bias—the first card is selected 43.4% of the time while the second is below random—raising discussion about placement fairness and algorithmic transparency.

4. Business model and policy implications: ReFiBuy's concept of agentic commerce optimization points to the emergence of related service models. Issues such as price mismatches and sponsored display slots also remind regulators to pay attention to information authenticity and fairness in AI shopping.