Claude and Claude Code Show Different Search Behaviours, Data Finds
Claude and Claude Code behave differently when answering the same prompts, according to new data from Profound, which monitors how brands appear in AI answers. Across 1,724 prompts tested between 13 and 23 July, Claude Code used web search in 13% of responses while Claude did so in 93%.
Brand mentions for the same prompt overlapped by about 20% on average. Repeat runs from the same product were more consistent than the two products were with each other. Two Claude responses to one prompt shared about half of their brands, while two Claude Code responses shared about 40%.
Claude Code mentioned nearly as many brands per response despite searching less often, 6.6 against Claude's 5.2. Its answers were shorter and more structured, averaging 322 words versus 459, and more than half included a table compared with 11% for Claude.
A second dataset covering the top 1,000 pages visited by each product's agent between 18 July and 18 August found that almost three-quarters of Claude Code's visits went to documentation, informational and pricing pages, against 5% for Claude. In contrast, 60% of Claude's visits were to robots.txt files, sitemaps and homepages, compared with 4% for Claude Code.
Profound advises brands to state technical and pricing facts plainly on pages Claude Code's agent visits, such as supported language versions, and to lead with the answer under question-shaped headings. The report does not test whether those changes lift brand mentions, and it does not explain how the two agents' traffic was distinguished.
Reported by Search Engine Journal ↗
Topics: Claude, Measurement, Citations