ChatGPT performs a search in nearly one-third of prompts

New data shows LLMS are acting like power search users, opening new ground for SEO strategy

A new study from marketing agency Nectiv reveals that ChatGPT triggers a web search in 31% of user prompts, averaging more than two searches per query — each far longer and more specific than typical Google searches. The findings suggest that SEOs can increasingly shape what ChatGPT retrieves and presents, particularly for commercial and local-intent queries.

Key takeaways

• Frequent search behavior: ChatGPT performed at least one search in 31% of prompts, averaging 2.17 searches per query.
• Longer and richer queries: The model’s searches averaged 5.48 words — about 60% longer than typical Google queries, with 77% being five words or more.
• Commercial and local focus: ChatGPT leaned most heavily on external searches for local-intent content (59%), and least for credit cards (18%) and fashion (19%).
• Dominant content types: Review-based, fresh, and comparative terms like reviews, features, and comparison were among the most frequent search modifiers.
• SEO leverage point: Because ChatGPT’s search layer pulls structured, commercial-style data, SEOs can influence its responses by optimizing for clarity, relevance, and freshness.
• Data scope: The study analyzed over 8,500 ChatGPT prompts across nine industries, generating 2,600+ search queries to track search frequency, depth, and intent.

Get the full story at Nectiv