The content types most cited by LLMs

We wanted to better understand which content types are most likely to be cited by different LLMs. To answer this question, we analyzed 75,000 AI answers, with 1,056,727 citations retrieved from three AI models (ChatGPT, Google AI Mode, and Perplexity).

We hypothesized that the ideal content type could vary by intent, so we segmented across four prompt intents.

1. Informational: The user is seeking knowledge, answers, or learning about a topic. They want to understand something better but aren't necessarily looking to make a purchase or visit a specific site. For example, "what is blockchain?" and "what are the symptoms of flu?"
   

2. Commercial: The user is researching products or services with purchase intent, but hasn't decided what to buy yet. They're comparing options, reading reviews, or looking for recommendations before making a decision. For example, "best smartphones under $500" and "top coffee makers 2026."
   

3. Navigational/local: The user wants to reach a specific website, page, or online destination, or they want to find a local service. They know where or what local service they require.  For example, "marketing consultant near me."
   

4. Transactional: The user is ready to complete an action or purchase right now. They've made their decision and want to buy, download, sign up, or execute a specific action immediately. For example, "book flight to Paris" and “I need to order running shoes online today.” 
   

Note: In reality, transactional prompts may often be branded since the consumer is often aware of the brand they wish to purchase. However, for the methodology in this study, it was important to keep all prompts non-branded.  

Then, to see if industry played a role in our findings, we further segmented across five industries:

1. eCommerce

2. Health & wellness

3. Home repair

4. Professional services

5. SaaS

   
   

All data was created and retrieved in the Peec AI platform, where I work as a GEO researcher focused on decoding how LLMs work and rank information.

The findings

I’ll give you a quick spoiler: Query intent was more predictive of content type citation than both industry and model choice, suggesting that understanding what users are trying to accomplish matters more than what industry you're in, or what model you are optimizing for. That said, there are many subtle differences that I’ll explore more deeply next.

Overall AI citations by content type 

Across all intents and verticals, listicles (21.9%), articles (16.7%), and product pages (13.7%) are the most common website citations, making up more than half of all citations. Product alternative and comparison pages pull in less than 3% of citations across all verticals.

Listicles have come under fire recently as Google seems to be cracking down on brands that use listicles for self-promotion. But we need to make an important distinction between: 

1. Self promotional listicles where the same brand is placing itself number one in the list 

2. External listicles where a neutral third-party site is ranking a collection of brands

In the professional services industry, where the listicle citation rate is the highest, we looked at the top 1,000 URLs in terms of number of citations. From this, we calculated the percentage of self-promotional listicles compared to third-party listicles.

• Self-promo: 19.1% of citations

• Third-party: 80.9% of citations 

Cleary, it’s third-party listicles that are moving the needle in AI search. (Here's more on how to get featured in one from a product review editor.)

Top cited content types by intent

User intent is the strongest predictor of which content types get cited. Below, we break down the top performing content types for each of the four intent categories: informational, navigational, transactional, and commercial. You can also see the full breakdown in the chart here.

Content type by prompt intent (all models)

As you’ll see next, listicles and articles make up 67% of the content for informational queries. For navigational queries, 40% of surfaced pages were either product or category pages. Product and category pages also pulled in 40% of  transactional pages. While listicles and category pages make up over 52% of citations for commercial intent.

Informational intent

When users want to learn about a topic like "What are the latest trends in AI translation services," AI models overwhelmingly favor articles and listicles. Products, categories, and homepages are almost never used for informational queries.

Top content types:

• Articles: 45.48% (172.7% above average)

• Listicles: 21.68% (on par with average)

• How-to guides: 9.21% (48.3% above average)

Least used:

• Product pages: 3.45%

• Category pages: 1.74%

• Homepages: 0.42%

Navigational/local intent

When users know what they want, like "BI architecture consultant in my local area," LLMs prefer to cite product pages and category pages. Homepages and profiles become critical navigation hubs, cited far more often than average.

Top content types:

• Product pages: 21.95% (60.7% above average)

• Category pages: 18.31% (62.8% above average)

• Homepages: 13.56% (157.8% above average)
  

Least used:

• Listicles: 5.36%

• Articles: 3.54%

• Alternatives: 0.03%

  
  

Transactional intent

When users are ready to act on queries like "Most affordable designer perfume reviews," they go directly to product pages and category pages. Articles and listicles are almost never used for transactions because users are ready to buy, not research.

Top content types: 

• Product pages: 24.88% (82.1% above average) 

• Category pages: 14.97% (33.1% above average) 

• Homepages: 7.38% (40.3% above average)
  

Least used: 

• Articles: 5.58% 

• Comparisons: 1.5% 

• Discussions: 6.68%

Commercial intent

When users are comparing options with queries like "Where can I buy wireless earbuds with noise cancelling and free shipping?," listicles dominate. Users want structured comparisons and peer opinions, not detailed articles.

Top content types: 

• Listicles: 40.86% (86.7% above average) 

• Category pages: 12.42% (slightly above average) 

• Discussions: 11.44% (52.1% above average)
  

Least used: 

• Articles: 6.15% 

• Product pages: 7.14% 

• Homepages: 1.7%

  
  

Which content type does each AI model prefer?

All three models agree on listicles as first, but diverge significantly on the second most used content type. ChatGPT and Google AI prefer articles; Perplexity uniquely elevates discussions—like those on Reddit, Linkedin, and G2—to second place.

ChatGPT prefers informational content

The ChatGPT citation profile is characterized by an information-heavy distribution drawn from diverse sources. It’s noted as the most information-focused model, showing a strong preference for articles (long-form authoritative content) and having the highest article representation among the models studied.

Key characteristics:

• 43.32% of citations from articles and listicles

• Highest article representation among models (4.38% above average)

• Low discussion preference (-4.32%)

Google AI Mode has a wide citation distribution

The Google AI model citation profile is characterized as having the most balanced and comprehensive citation distribution. It represents all 11 content types with minimal bias and shows a strong performance across all intent types, suggesting a balanced approach with a slight preference for educational content such as articles and how-to guides.

Key characteristics:

• Articles and how-to guides slightly above average (+2.14% and +1.58% respectively)

• Discussions are less represented (-5.51%)

Perplexity citations lean toward discussion pages

The Perplexity citation profile is characterized by its community and discussion-focused distribution. It shows a strong preference for discussions and consequently de-emphasizes traditional articles. This suggests the model values community opinions and peer insights.

Key characteristics:

• 17.35% of citations from discussions, more than double the 7.52% average

• Lowest article usage among models (-5.0%)

• Most egalitarian intent distribution (~25% each)

How citations vary by industry

While the data reveals clear content preferences by industry, prompt intent is a much stronger driving force in what content types LLMs prefer. However, analyzing your niche as well as query intent can help you get the most accurate insights.

• SaaS shows the strongest bias toward listicles, reflecting the sector's comparison-driven research behavior, though articles still maintain a moderate presence. Professional services follows a similar pattern but less dramatically.

• Health & wellness inverts this trend, with articles outpacing listicles, emphasizing the need for authoritative, educational content in this trust-sensitive sector.

• eCommerce demonstrates the most balanced distribution across listicles, articles, and category pages, indicating multiple viable content strategies work equally well.

• Home repair shows the flattest distribution across content types, with no single format dominating, suggesting AI search pulls from diverse content sources for service-based queries.

• Notably, profile and discussion formats remain minimal across all industries, indicating AI citations currently favor structured editorial content over individual business pages or community discussions. 

The takeaways: Understanding the prompt intent you’re optimizing for is critical for your content strategy. From there, you can analyze the top-content types for your specific niche to make your focus even more effective. 

eCommerce

Top content types for eCommerce AI citations are:

1. Listicles (19.94%) - Product comparisons and roundups

2. Articles (19.49%) - Product reviews and guides

3. Category pages (15.96%) - Product browsing

Health & wellness 

Top content types for health and wellness AI citations are:

1. Articles (19.66%) - Medical credibility and authority

2. Product pages (14.77%) - Supplement and wellness products

3. Other (14.24%) - Diverse health topics

Home repair 

Top content types for home repair AI citations are:

1. Product pages (18.52%) - Tools and materials

2. Category pages (14.95%) - Product browsing

3. Articles (12.48%) - DIY guides and information
   

Professional services 

Top content types for professional services AI citations are:

1. Listicles (25.24%) - Strongest listicle citations of any industry

2. Articles (16.8%) - Thought leadership and expertise

3. Product pages (13.12%) - Service offerings

4. Category pages (10.06%) - Service categorization
   

SaaS 

Top content types for SaaS services AI citations are:

1. Listicles (35.37%) - Tool comparisons and reviews

2. Articles (16.33%) - Technical documentation and education

3. Product pages (12.75%) - Product pages and pricing

4. Category pages (6.97%) - Service categorization

Critical takeaways for your AI content strategy

The central finding: query intent matters more than industry or model.

Understanding what users are trying to accomplish matters more to your AI search content strategy than what vertical you're in or which AI model you're optimizing for. A commercial query in health follows similar citation patterns to a commercial query in SaaS, despite these industries having different content needs overall.

The strategic implication: The sharp divergence between research content (articles and listicles) and conversion content (product and category pages) means brands must build distinct content ecosystems for different stages of the user journey. Articles build trust but don't drive decisions. Listicles drive commercial research but disappear during transactions. Product pages convert but don't educate.

Model differences exist but are overrated: While ChatGPT skews toward articles, Google AI maintains balance across all content types, and Perplexity uniquely elevates discussions, these differences are less significant than intent-based patterns. For most content strategies, optimizing for user intent will outperform optimizing for specific AI models. The exception: Discussion and review content is more critical if your audience heavily uses Perplexity.

Optimization recommendations: a recap

For informational prompts

✓ Prioritize long-form articles ✓ Use how-to guides for procedural information ✓ Add listicles for curated information ✗ Avoid product pages and category pages

For commercial prompts

✓ Lean on listicles ✓ Include discussion/review content ✓ Use category pages for browsing ✗ Don't rely on articles ✗ Avoid product pages in research phase

For navigational/local prompts

✓ Ensure product pages are discoverable ✓ Maintain strong category page hierarchy ✓ Make sure you have accessible about pages and clear homepage navigation ✗ Don't focus on listicles

For transactional prompts

✓ Optimize your product pages ✓ Have clear category pages for browsing ✓ Have transaction pathways on your homepage ✓ Publish how-to guides for complex products ✗ Avoid long-form articles

Industry-specific optimization

• eCommerce: Balance articles with category pages

• Health & wellness: Invest heavily in authoritative articles

• Home repair: Focus on product pages and how-to guides

• Professional services: Publish listicles for comparisons and articles for authority

• SaaS: Make sure your content strategy includes listicles and comparison/review content