A recent study by Havas Market, published on Thursday, July 17, highlights the overwhelming presence of brand-generated content in responses provided by Google’s AI assistant, Gemini. Based on an analysis of 11,000 responses to 150 common user questions across twelve sectors — ranging from health and fashion to finance and automotive — the report reveals that 57% of answers in France originate from manufacturer or service provider websites.
According to Havas Market, this dominance is largely driven by the abundance of brand-produced materials, including technical product sheets, FAQs, corporate blogs, and instructional guides. In contrast, content from public institutions, media outlets, and retailers appears far less frequently in the AI’s answers.
The numbers are clear: 57% of Gemini’s responses reference corporate sources, while only 16% come from public institutions, 15% from media, and just 12% from retailers. Prominent brands such as Apple, Renault, and L’Oréal frequently appear among the top referenced sources.
This trend is especially pronounced in certain industries. Sectors like fashion (75%), DIY (74%), finance (75%), technology (65%), and automotive (60%) are overwhelmingly dominated by brand-generated content. Meanwhile, more regulated areas like health (49% institutional sources) and energy (26%) see a stronger presence from public institutions.
Maxime Guernion, SEO expert at Havas Market, explains, “Brands have been investing heavily in high-quality, SEO-optimized content for years, making it easier for AI systems like Gemini to select their content as answers to user queries.”
The study also points to the critical role of technical website performance, including fast load speeds, mobile optimization, and AI-friendly content structures. These factors align with Google’s recommendations, which prioritize original, informative content with structured data and high-quality multimedia elements.
Public institutions, however, struggle to achieve the same level of visibility. A few key players still stand out in specific sectors — Ameli.fr in health, SNCF Connect in travel, Leroy Merlin and Castorama in DIY, and general knowledge platforms like Wikipedia and YouTube.
Notably, 53% of Gemini’s responses draw from sources that are repeatedly cited, indicating a high level of concentration around a limited number of dominant sites. Nevertheless, niche websites and smaller businesses can still break through if they produce useful, specialized, and well-optimized content.
The findings raise important questions about the neutrality of AI-generated information. While AI tools like Gemini aim to deliver accurate and high-quality responses, the disproportionate reliance on brand-driven content raises concerns about objectivity, as these sources naturally promote their own products and services.
