New study: Where ChatGPT already works in online shopping – and where it does not

• Analysis of 973 e-commerce websites with more than USD 20 billion in revenue shows ChatGPT’s limited role in online shopping
• For complex products with high information needs, ChatGPT referrals can create value
• For standard products involving simple decision-making processes, ChatGPT lags behind most channels

ChatGPT was quickly framed as a potential “Google killer” in online shopping. A new study by Frankfurt School of Finance & Management paints a more nuanced picture: for complex purchase decisions, ChatGPT can already be a relevant channel. Across broader e-commerce, however, its actual role remains much smaller than expected.

Professor Christian Schulze, Professor of Marketing at Frankfurt School, and Maximilian Kaiser from the University of Hamburg examined organic traffic from ChatGPT to e-commerce websites and compared it with established digital marketing channels such as organic search, paid search, email marketing, affiliate marketing and paid social. The analysis is based on data from 973 e-commerce websites with combined revenue of more than USD 20 billion during the study period from August 2024 to July 2025.

Value for complex purchase decisions

One year after the launch of organic shopping links, traffic from Large Language Models accounts for less than 0.2 per cent of total traffic in the sample studied. ChatGPT also does not yet match most traditional channels on key performance indicators. Its conversion rate and revenue per session are higher than that of paid social, but lower than all other established channels. At the same time, the data show a positive trend: ChatGPT conversion rates are increasing over time.

For retailers, this means that generative AI is not yet a substitute for established channels. In information-intensive categories, however, ChatGPT can already be a useful complement, particularly where consumers seek guidance before making a purchase decision. For providers of simple products, the short-term value remains limited.

For LLM platforms, the study shows that the next stage of development lies in better product availability, more reliable recommendations and a smoother purchasing process. Since the end of the observation period, LLM platforms have launched several initiatives to address this limitation, including instant checkout and agentic shopping. 

“Our findings contradict the idea that ChatGPT has already transformed online shopping — but the real story is in the product categories. The study shows where ChatGPT already has substance in e-commerce and where it remains largely hype,” says Professor Christian Schulze. “For simple purchase decisions, its value is currently limited. For more complex products, however, ChatGPT can already create value by structuring information, making alternatives comparable and supporting consumers in their decision-making process. Companies should therefore assess very carefully whether their products benefit from this type of guidance.”

The study “ChatGPT Referrals to E-Commerce Websites: How Do LLMs Compare Against Traditional Channels?” is available in Marketing Science.