NEW YORK, NY — As retailers gear up for the holiday shopping season, a new study from Columbia Business School shows how artificial intelligence is already shaping what shoppers see online—and how they decide what to buy. The research finds that Generative AI (GenAI) makes online shopping noticeably smoother by helping people find relevant products faster, understand what they’re purchasing, and move through the buying process with greater confidence.
In the paper, Generative AI and Firm Productivity: Field Experiments in Online Retail, Columbia Business School Professors Dante Donati and Miklos Sarvary, along with their co-authors, analyzed seven major uses of AI on a global online retail platform to understand how these tools shape the way people shop. By testing AI-powered features like smarter search, pre-sale chatbots, and richer product descriptions through large-scale experiments, the researchers found that AI consistently made the shopping process easier — helping people get answers faster, understand what they were looking at, and find products that better matched their needs. These improvements led more shoppers to complete their purchases, resulting in clear, measurable sales lifts.
“This study matters because it shows that AI can improve online shopping in practical, meaningful ways for both retailers and consumers,” said Dante Donati, Assistant Professor of Business at Columbia Business School. “For retailers, small upgrades to search or customer support can translate into more completed purchases. And for consumers, AI makes the process clearer, simpler, and more trustworthy.”
For the study, Donati and Sarvary, together with co-authors Lu Fang of Zhejiang University, Zhe Yuan of Zhejiang University, and Kaifu Zhang of the Chinese University of Hong Kong (Shenzhen), conducted large-scale randomized field experiments between September 2023 and June 2024, involving millions of shoppers across one of the world’s largest online marketplaces. In each test, some users saw the platform’s traditional version of a feature—such as search, customer service chat, or product descriptions—while others saw a new version enhanced with Generative AI. By comparing how the two groups behaved in real time, the researchers could see exactly how AI changed what people clicked on, which products they viewed, whether they completed their purchases, and how much they spent. Because prices, inventory, and staffing remained constant, the team was able to cleanly isolate AI’s impact on the shopping experience and pinpoint where the technology made the greatest difference.
The study also finds that AI is especially helpful for people who may have a harder time shopping online today. Newer shoppers, people using the platform less frequently, and smaller or newer sellers all benefited more when AI tools were switched on—suggesting that AI has the potential to make digital shopping easier and more inclusive.
Key findings from the research include:
- AI improves shopping without changing prices or staffing: The sales lift came entirely from a smoother, more helpful shopping experience — not from discounts, promotions, or operational changes.
- AI makes it easier for people to find what they actually want: Better search, clearer product information, and faster answers help shoppers find the right products more often, which drives more completed purchases.
- The benefits stack across different parts of the shopping journey: Each AI upgrade — search, chat, product descriptions — added its own small improvement, and together they created a noticeable boost in overall performance.
“As AI continues to catch on, business leaders should invest in tools that truly help people make sense of their options,” said Miklos Sarvary, Carson Family Professor of Business at Columbia Business School. “The retailers who win will be the ones who use AI to build trust and make shopping easier—not more confusing.”
