AI Models Favor Familiar Brands in Search: Study

Artificial intelligence models demonstrate a significant preference for searching familiar brands, doing so 3.2 times more frequently than unfamiliar ones, according to a recent study by geoSurge. The research revealed that models accessed brands they already knew 55.7% of the time, in stark contrast to the 17.4% of the time they searched for brands outside their top ten most familiar. This bias appears to be strongly influenced by the model's internal memory, which researchers measured independently to assess its impact on search behavior. Across nearly 4,000 responses to 66 buyer-focused questions originating from the U.S., the consistent pattern emerged: familiar brands were prioritized in search queries. Further analysis showed that when models did conduct "fan-out" searches, which explore related information, the majority did not explicitly name a brand, with only 31% of these searches including a company name. When a specific brand was named, a substantial 63% of those instances involved one of the model's top five most familiar brands. The study also examined this phenomenon across various industries, including travel, automotive, finance, business software, education, food and restaurants, luxury goods, fitness and wellness, and fashion. Within these sectors, models searched for familiar brands between 41% and 82% of the time, while unfamiliar brands were accessed only 9% to 23% of the time. Researchers acknowledged that the sample size for some industries was limited, with as few as six prompts used for certain sectors. While memory was a primary driver, it did not dictate every search outcome. An example cited involved Google's Gemini model searching for "Lemon Squeezy" when answering a query about online payment providers, despite this brand not being present in its measured memory. This suggests that live search capabilities can still uncover less familiar brands, particularly in categories where models rely less on pre-existing knowledge. The implications of these findings are significant for brands. The study indicates that established brands may possess an inherent advantage in AI-driven search results, potentially influencing consumer discovery before a search even begins. While compelling content can still help newer brands gain visibility, the research points to a potential "unfair advantage" for those already well-known to AI models. The data was collected between May 29 and June 9, involving 66 U.S. buyer prompts tested 60 times each, resulting in the analysis of 3,960 responses, 13,281 fan-out searches, and 1,416 brand-level observations. The authors concluded that a clear relationship exists between a model's memory and its subsequent search behavior, highlighting the importance of brand recognition in the evolving landscape of AI-powered information retrieval.