Large language models (LLMs) often scour the web to supplement their responses with the latest and most relevant information to users’ prompts. Each model cites different domains to different degrees, depending on the crawling permissions publishers give them, the models’ training and other factors.
But after reviewing more than 40,000 URLs that ChatGPT cited heavily on the Evertune platform in the last 60 days, we found a preference much stronger than its preference for any particular domain: ChatGPT loves listicles.
About 50% of the URLs we examined overall were listicles. That was also true for the majority of the nearly 50 product categories within our sample.
Of the more than 21,000 listicles we reviewed, 58% were ranked lists, like “10 Best CRM Tools,” for example. Another 17% were unranked lists; think “7 Great Gifts For…” or similar. Institutional rankings, data-heavy lists like U.S. News & World Report’s Best Colleges rankings, constituted 11%. Only 15% were some other type of listicle that didn’t fall into one of these three categories.
Statista, Forbes and Wikipedia were the most-cited domains of these listicles.
Given that Evertune is an AI marketing platform for brands to track and improve their AI visibility, it is not altogether surprising that listicles are so heavily cited. Brands want to know what models are saying about their product category and how that compares to their competitors. That is often exactly what listicles do: compare top brands and products side-by-side. Listicles also include information vital for the product comparison tables ChatGPT now features in its shopping widget.
Furthermore, listicles follow some Generative Engine Optimization (GEO) best practices for getting content cited by AI models, namely:
- Comparison tables
- Numbered headers
- Highly structured content
- Deep specificity into a single topic
All of these factors make for content that is highly digestible for LLMs.
Other Page Components
Of the more than 40,000 links that ChatGPT cited heavily within the last 60 days, we were able to review the page structures for more than 33,000 URLs.
No page element jumped out as a key ChatGPT preference to the degree that listicles had, but there were points of consistency for most of these URLs. For example, pages typically had about 1,000 words, 18 paragraphs and 10 images.
After removing outliers, the middle 50% of the URLs (the data within the 25th and 75th percentiles) were as follows:
| Page Component | Middle 50% Range | Median |
|---|---|---|
| Words | 127–1,818 | 941 |
| Paragraphs | 2–40 | 18 |
| Avg. Sentence Length | 12–21 | 17 |
| H2s | 0–9 | 4 |
| H3s | 0–9 | 2 |
| Internal Links | 0–86 | 28 |
| External Links | 2–42 | 15 |
| Images | 1–22 | 10 |
| Lists | 0–8 | 2 |
Keep these in mind when considering how to optimize content for AI consumption.
Methodology
At Evertune Research, we track hundreds of brands across 250 categories. For this analysis, we reviewed more than 40,000 URLs that ChatGPT cited heavily in the last 60 days across nearly 50 product categories.
Evertune is the AI marketing platform for Generative Engine Optimization (GEO) that helps brands improve visibility in AI search by analyzing responses at scale and delivering actionable insights. Evertune works with leading brands across all verticals, including Finance, Retail and E-Commerce, Automotive, Pharma, Tech, Travel, Food and Beverage, Entertainment, CPG, and B2B. Founded by early team members of The Trade Desk, Evertune has raised $20M in funding from leading adtech and martech investors. Headquartered in New York City, the company has a growing team of more than 40 employees.
Don't forget to share this post
