**A new study shows that self-promotional “best” lists play a bigger role in AI recommendations than most people realize. When these lists are updated and clearly structured, they are far more likely to appear in ChatGPT’s answers.**

[Ahrefs](https://ahrefs.com/blog/best-lists-research/) researcher Glen Allsopp reviewed more than 26,000 URLs cited in ChatGPT’s responses and found a consistent pattern. 

When users ask for software suggestions, product options, or agency recommendations, ChatGPT often relies on comparison-style blog posts. 

His team analyzed 750 top-of-funnel prompts and saw these lists appear again and again.

The research, completed in December 2025, compared ChatGPT’s sources with trends seen across other AI tools. The goal was to identify which types of pages show up most often when AI systems provide recommendations.

The findings drew quick reactions from industry leaders as well. Rand Fishkin noted that the research highlights how easily brands can push their way into AI-generated answers through heavily self-promotional “best of” lists, raising questions about how AI tools evaluate quality and intent.

 

> If you think AI tools are ethically shady, you have a moral duty to check out [@ahrefs](https://twitter.com/ahrefs?ref_src=twsrc%5Etfw) research from last week showing how to effectively spam your way into AI mentions with self-promotional “best of” lists 👇 [pic.twitter.com/7V4wIPNM7h](https://t.co/7V4wIPNM7h)
> — Rand Fishkin (follow @randderuiter on Threads) (@randfish) [December 12, 2025](https://twitter.com/randfish/status/1999392298925429010?ref_src=twsrc%5Etfw)

 

## What Stood Out in the Findings

Here are some of the most notable patterns the research revealed.

- ### “Best” lists appeared again and again

The most consistent trend was the dominance of listicle-style blog posts. Out of all page types identified, almost 44% were “best” lists. That meant a large share of ChatGPT’s recommendations were shaped by pages that ranked or compared several brands. This result held across every niche analyzed in the dataset.

![How prominent "best" blog lists are - AI Favors best lists](https://www.stanventures.com/news/wp-content/uploads/2025/12/top-of-funnel-best-lists-300x300.webp)

- ### Fresh updates seemed to matter a great deal

Allsopp’s team noted that many of the lists pulled into ChatGPT answers were updated recently. Roughly 79% had been updated in 2025, and a significant portion had been modified within the two months leading up to the study. 

![Cited List Dates](https://www.stanventures.com/news/wp-content/uploads/2025/12/cited-list-dates-300x291.webp)

This pattern suggested that ChatGPT may favor pages that signal freshness, even when the ranking itself is self-promotional.

- ### List position influenced visibility

The study also observed a pattern relating to placement. Brands listed near the top of a comparison article tended to appear more frequently in ChatGPT’s recommendations. This correlation showed up regardless of how long the list was, indicating that both ChatGPT and other AI tools may pay attention to ranking order in these pages.

- ### Low-authority sites still made it into AI answers

One of the more unexpected findings was the presence of low-authority publishers. Some cited sites had minimal organic traffic and few external signals of quality. Despite that, their “best” lists were still referenced by ChatGPT. This raised questions about long-term reliability and how AI assistants weigh the credibility of sources.

- ### Different niches leaned on different page types

Although list posts were the top category across the board, the breakdown by niche revealed more nuance.

**Software**: Landing pages and documentation also appeared frequently.

**Products**: “Best X” list posts dominated by an even wider margin.

**Agencies**: List articles and agency landing pages carried most of the weight.

These distinctions showed that AI systems do not use a single content pattern for every industry. They pull

## What This Means for Companies Hoping to Be Recommended by AI

These findings offer two clear insights. Companies can improve their visibility by producing helpful comparison pages. 

At the same time, they need to be careful not to publish content that feels exaggerated or one-sided. 

AI assistants respond well to pages that explain real differences between competing products and give readers practical information that supports decision-making.

## How Brands Can Increase Their Chances of Being Cited

Here are practical steps companies can take.

1. Create comparison pages that help readers understand real options.
2. Keep those pages updated. Include clear dates and review content when products change.
3. Be transparent about alternatives. Explain differences in features, pricing, and ideal users.
4. Focus on clarity. A few strong pages work better than many thin ones.
5. Pay attention to engagement. Look at click patterns and time on page to understand what readers find useful.

Some companies choose to work with [AI SEO](https://www.stanventures.com/ai-seo-services/) specialists who study how assistants select sources. Professional guidance can help identify content gaps, improve structure, and ensure that comparison pages give AI systems the clarity they need

## What AI Systems Seem to Prefer

Three factors appeared consistently: freshness, clarity, and ranking position within a list. Landing pages and documentation also played an important role, especially in software and agency recommendations.

These signals suggest that companies should maintain well-structured product pages while also publishing comparison posts that place their product within a larger set of options.

## What to Watch Going Forward

There are still risks. Working with low-quality list sites can weaken trust. Over-optimizing for AI can lead to weak content that frustrates readers. Strong performance usually comes from pages that help people make informed decisions instead of trying too hard to influence an algorithm.

## Key Takeaways

- Best” list articles make up a significant share of the sources used by AI assistants. 
- Recent updates improve the chances that a page will be used as a citation. 
- Higher list placement often results in more mentions in AI responses. 
- Some cited lists come from low-authority sites, which raises quality concerns. 
- Landing pages and documentation still matter for software and agency recommendations.