This paper examines how standalone large language models (LLMs), exemplified by ChatGPT, reshape consumer behavior in commercial search. We conceptualize a key economic cost in traditional search as articulation cost—the cognitive effort required to...
This paper examines how standalone large language models (LLMs), exemplified by ChatGPT, reshape consumer behavior in commercial search. We conceptualize a key economic cost in traditional search as articulation cost—the cognitive effort required to translate vague needs into effective search queries. Using Nielsen Korea panel data from November 2022 to April 2023, we estimate the causal effects of ChatGPT adoption. We find that users increase commercial search volume by 31.7% and search variety by 29.9%, shifts queries toward lower-purchase-funnel stages, and engage more with downstream e-commerce pages. These patterns suggest that LLMs reduce articulation cost, helping users continue search sequences. We contribute to strands of research. First, in marketing and economics, we formalize articulation cost as a distinct economic cost, showing that canonical search model assumptions—that consumers can readily translate needs into queries—may not hold. Second, in search advertising, we show how technology can reduce query ambiguity, complementing prior work on its impact on advertising and models for inferring user intent. Third, in research on generative AI, we demonstrate in an e-commerce context that users often reinvest the cognitive effort saved into exploratory search rather than merely completing tasks efficiently, extending the discussion of AI-driven effort allocation.