GEO Experiments Challenge AI Visibility Advice

Two structured experiments have been conducted to test conventional advice regarding generative engine optimization (GEO) for AI platforms, yielding measurable results that challenge existing theories. The first experiment, focused on an established brand, involved significant investment over several months and tracked 15 commercial-intent keywords across four AI platforms: ChatGPT, Claude, Gemini, and Perplexity. Queries were executed manually, with and without a VPN, to account for potential location-based variations. The strategy involved publishing listicles on sources frequently surfaced by Large Language Models (LLMs) for the target keywords, alongside content on the brand's own website. This effort resulted in the brand appearing for approximately 10 to 12 of the 15 keywords, with peak keyword presence reaching 37.01% on April 29. Citations by platform were recorded as follows: ChatGPT received 148, Claude 96, Gemini 87, and Perplexity 64. The most significant source of citations was a detailed listicle on Indeed SEO, which garnered 190 mentions, far exceeding other sources like MEXC and a GlobeNewswire release. Listicles constituted 72.4% of all citations, with PR contributing 24.1%, and guest posts, the owned site, and LinkedIn accounting for the remainder. Four key lessons were derived from this initial test.

The second experiment was a 30-day cold-start initiative for a SaaS link-building agency that had no prior measurable AI presence. This test also tracked 15 commercial-intent keywords but expanded the AI platform coverage to six. A total of 775 citation events were logged across both experiments. One of the initial conclusions drawn from the first experiment did not hold up after the second test was completed, indicating a need for re-evaluation of certain GEO strategies. The second experiment aimed to provide a clearer understanding of how to establish AI visibility from scratch, testing the efficacy of different approaches in a more controlled, short-term timeframe. The detailed logging of citation events and keyword performance across multiple platforms and over distinct periods allowed for a comparative analysis of what strategies are most effective in the evolving landscape of AI-driven search and content discovery. The findings from these experiments are intended to offer practical, data-driven insights for businesses and marketers seeking to improve their visibility within AI search environments, moving beyond theoretical recommendations to actionable, evidence-based tactics. The manual tracking and documentation process ensured a high degree of accuracy in the collected data, providing a robust foundation for the conclusions drawn. The experiments highlight the dynamic nature of AI visibility and the importance of continuous testing and adaptation of GEO strategies. The specific platforms tested, the methods of citation, and the keyword tracking provide a concrete framework for understanding the nuances of AI SEO. The results underscore that while certain tactics like listicles and PR remain effective, their precise impact and the overall strategy may need adjustment based on the specific AI platform and the competitive landscape. The disparity in citation volume across platforms also suggests a differential in how various LLMs surface and prioritize information, a critical factor for any GEO strategy targeting AI. The experiments aim to demystify the process of gaining visibility in AI-driven search results, offering a practical guide based on empirical evidence rather than speculation.