According to May data from Adobe Analytics, U.S. shoppers using large language models like Google's Gemini or OpenAI's ChatGPT for purchase recommendations linger longer on retailers' websites and are more likely to spend. Consumers referred from LLMs generated 53% more revenue per visit than shoppers from non-AI sources. AI traffic to retail websites increased 138% in May from last year, the highest share of total retail visits since Adobe Analytics began tracking in October 2024. Retail website visitors recommended by AI converted at a rate 54% higher than online shoppers from non-AI sources in May. Shoppers referred to e-commerce websites spent 53% more time on the sites than visitors from other sources, and AI-referred shoppers also visit more retail webpages than non-AI referred visitors. Vivek Pandya, director of digital insights at Adobe, said retailers whose products appear in LLM suggestions can "drive more personalization" to shoppers who leave the platforms to complete purchases on native websites.