Large language models (LLMs) are exceptionally good at distilling vast amounts of data into tailored answers to natural-language prompts. Their outputs rely on huge training corpora and are frequently refined with live internet searches. The quickest way to shape what LLMs say about your brand is to shape the content they surface in those searches. At Evertune Research, we rely on the Evertune AI marketing platform to monitor hundreds of brands in 250 categories across every major LLM. This lets us see which sources AI systems reference most frequently, especially when users request brand or product recommendations in different sectors.
For this study, we examined the 6,000 most-cited URLs per model for March and April across ChatGPT, Copilot, Gemini, Google AI Mode, Google AI Overview, and Perplexity. A clear pattern emerged: these systems lean heavily on listicles.
Half of LLMs’ top-cited URLs are listicles
Among the roughly 25,000 unique URLs we analyzed, about half were listicles. Looking at nearly 400 million citations across all models, 63% referenced listicles. Several traits make listicles especially attractive to LLMs. They typically focus on a narrow topic—such as “best laptops for gamers”—which aligns closely with user intent. Their structured, numbered format also makes them straightforward for models to interpret and reuse. For brand-focused questions, listicles effectively pre-compare products on attributes like features, price, and materials, mirroring the comparison style that ChatGPT now highlights in its shopping widget. Listicles appeared at scale across every model we examined…