Can You SEO Your Way into an AI Agent's Recommendation? (www.joe-shirey.com)

Recent observations reveal a significant shift in consumer behavior regarding product research, with 46% of AI users now turning to standalone AI platforms like ChatGPT for recommendations, doubling from just 25% in 2024. This trend presents a challenge for SEO strategies, as vendors ponder if they can influence AI agents' recommendations through manipulated search results. A study conducted by Armature with the coding agent Claude Code showed that while AI tools frequently access search engines, the dependency on traditional SEO techniques falls short of affecting real-time recommendations, as seen in a series of controlled tests. The research highlights that while AI agents predominantly rely on their training data for recommendations, prompting them to search can significantly alter the results. In scenarios where agents were encouraged to search, the likelihood of mentioning a Google Cloud product soared from 9% to 17%, showcasing a ceiling effect of manipulated search inputs. Despite this, the primary learning takeaway indicates that effective recommendations stem more from the agents' training rather than current SEO efforts. Developers are encouraged to proactively ask AI agents to verify current documentation to ensure relevant and updated suggestions, rather than relying solely on SEO tactics implemented by vendors.