Analysis/Study | Optimization for ChatGPT: Why LLMs.txt hardly matters, but brand mentions on Quora and Reddit are crucial

The DNA of AI responses decoded: Why classic authority and social proof beat technical gimmicks

For a long time, optimizing for Large Language Models (LLMs) like ChatGPT was like gazing into a crystal ball. SEO experts and marketing professionals relied on guesswork, hope, and technical experiments like the `LLMs.txt` file or special schema markup to appear in the AI's generated responses. But a new, comprehensive data analysis now dispels these myths and provides, for the first time, reliable facts about what really motivates ChatGPT to cite a source.

This study, based on the analysis of 129,000 domains and over 216,000 pages from 20 different industries, reveals a clear pattern: AI cannot be tricked by technical "hacks." Instead, it confirms a return to the fundamental values ​​of the web—trust, authority, and genuine human interaction. While much-discussed technical approaches, such as including an `LLMs.txt` file, actually worsened the predictive accuracy of the analysis, two powerful levers emerged: massive domain authority and a presence on discussion platforms.

Suitable for:

• The study comes from our Business Partner SE Ranking: “ How to Optimize for ChatGPT: Skip LLMs.txt, Earn Trust on Quora & Reddit ”

Particularly noteworthy is the resurgence of forums and communities. The data clearly shows that brand mentions on platforms like Quora and Reddit act as a crucial trust signal – domains with a strong presence on these networks are four times more likely to be cited by AI. At the same time, “content is king,” albeit under new circumstances: ChatGPT favors extremely in-depth content of over 2,900 words and drastically rewards recency – those who update their content quarterly can double their visibility.

This article summarizes the key findings of the study and shows why the path to AI-driven responses lies not in new file formats, but in building digital authority and genuine community relevance. Learn in detail which factors maximize your chances and which optimizations you can skip.

Suitable for:

• Experts in Perplexity SEO or GEO: How to make your website a favorite of AI like Perplexity, Gemini, ChatGPT & Co.

What are the most important findings from current research on ChatGPT optimization?

A comprehensive study analyzed 129,000 unique domains and 216,524 pages across 20 different industries to understand the patterns in ChatGPT citations. The findings debunk some common myths and reveal what truly drives AI citations. Many bold claims circulate online about what prompts ChatGPT to cite specific sources. Some swear by LLMs.txt, others by domain authority, content recency, or even structured metadata like schema markup. This study separates fact from speculation and reveals surprising results.

What role does domain authority play in ChatGPT citations?

Domain authority proves to be one of the strongest factors for ChatGPT citations. Websites with over 32,000 referring domains are 3.5 times more likely to be cited by ChatGPT compared to websites with only up to 200 referring domains. This pattern also holds true for domain trust: Highly trusted domains with a DT score above 90 receive almost four times more citations than low-trust websites with a DT score below 43.

The largest growth spike occurs when exceeding the 32,000 backlink mark, where citations nearly double from 2.9 to 5.6. This clearly demonstrates that link authority accumulates. Once a website reaches a critical mass of backlinks, its perceived trustworthiness increases exponentially. The analysis also shows that linking from one's own website to other highly authoritative websites has minimal impact. Whether linking to domains with a Trust score of 70 or 100, the influence is virtually zero compared to inbound backlinks.

How important is domain trust for the likelihood of a citation?

A domain's trustworthiness has a significant impact on citation growth. Websites with a Domain Trust score below 43 struggle to gain traction, averaging only 1.6 citations. However, once a website's DT score reaches 77, noticeable benefits begin. The real acceleration starts at a score of 90, where citation growth becomes exponential and significantly faster.

Interestingly, having a trusted domain zone doesn't automatically guarantee higher citations. While one might expect domains with extensions like .gov or .edu to perform better than others, the analysis shows they achieve an average of about 3.2 citations. This is even lower than the 4 citations recorded for websites outside these zones. Ultimately, it's not the domain name itself that matters, but the quality of the content and the added value it provides.

What is the significance of Page Trust compared to Domain Trust?

Similar to domain trust, page-level trust is also crucial. URLs with a Page Trust Score above 23 begin to see tangible results, and those reaching 28 or higher consistently achieve an average of 8.2 citations. While a higher Page Trust Score can increase a page's chances of being cited by ChatGPT, it doesn't need to be as high as the Domain Trust Score.

Surprisingly, all pages with a score of 28 or higher are cited roughly the same number of times. This shows that ChatGPT is more interested in the overall authority of a domain than the trustworthiness of individual pages. The implication is clear: Building strong domain authority is more important than optimizing individual pages for maximum trust.

How does visibility on Google affect ChatGPT citations?

Beyond traditional authority factors, a website's overall visibility on Google, measured by traffic and average ranking, significantly influences its chances of being cited by ChatGPT. Domain traffic ranks second in importance for ChatGPT citations but remains less influential than backlinks.

Websites with fewer than 190,000 monthly visitors achieve an average of 2 to 2.9 citations. Only after exceeding the 190,000 visitor mark does a clear correlation emerge. Domains with over 10 million visitors achieve an average of 8.5 citations. This pattern suggests that large language models like ChatGPT are less concerned with popularity until it becomes undeniable.

Websites with low to medium traffic don't have a significant advantage in ChatGPT citations. For example, a website with 20 organic visitors and another with 20,000 visitors tend to receive roughly the same ChatGPT score. In such cases, other factors like content quality, relevance, and authority likely outweigh pure traffic numbers, giving even smaller websites a chance to be cited by ChatGPT.

Why is traffic to the homepage particularly important?

The analysis shows that it's not just about driving traffic to any page on the website. It's the global traffic to the homepage that counts. Websites with at least 7,900 organic visitors to their homepage have the highest chance of being cited by ChatGPT. This underscores the importance of a strong homepage as a key element of website authority.

Furthermore, the average position of a URL in organic Google search correlates with ChatGPT citations: Pages with an average ranking between 1 and 45 receive an average of 5 citations, while pages with a ranking between 64 and 75 see only 3.1 citations. Although this does not prove that ChatGPT relies on Google's index, it suggests that both systems evaluate authority and content quality similarly.

What influence do brand mentions on Quora and Reddit have on ChatGPT citations?

Today, a presence on discussion platforms acts like digital word-of-mouth. Domains with millions of brand mentions on Quora and Reddit are roughly four times more likely to be cited than those with minimal activity. On Quora, websites with minimal presence (up to 33 mentions) receive an average of 1.7 citations, while websites with a strong Quora presence (6.6 million mentions) receive an average of 7.0 citations.

The same pattern holds true for Reddit: citations range from 1.8 to 7 across the entire scale. This is particularly encouraging for smaller, less established websites: engagement on Quora and Reddit gives them the opportunity to gain trust in ChatGPT's eyes. In other words, it's a way to build authority, much like established domains achieve through diverse referring domains and high web traffic.

The most important practical steps for optimizing for ChatGPT include participating in relevant discussions, not just advertising, encouraging organic brand mentions through helpful contributions, and using these platforms to demonstrate expertise and authority.

How important is comprehensive and in-depth content for ChatGPT citations?

Length and depth correlate directly with higher ChatGPT citations. Short articles with fewer than 800 words average 3.2 citations, while long-form posts with over 2,900 words achieve 5.1 citations. However, the key is not length for its own sake, but depth. ChatGPT favors pages that capture the full context, nuances, and subtopics of a subject.

For most topics, a minimum of 1,900 words is recommended. If necessary, the word count should be increased to at least 2,900 words. It is important to cover related concepts, synonyms, and examples to improve semantic diversity. Supporting data such as expert citations and statistics should be included.

The analysis shows that pages with expert citations achieve an average of 4.1 citations, compared to 2.4 without. Pages rich in statistics (19 or more data points) achieve an average of 5.4 citations, compared to 2.8 with minimal data. However, these factors have a relatively small impact compared to others and should be considered supporting indicators of high-quality content.

What role does content structure play in thematic clarity?

Even the best content can underperform if it's structurally opaque. ChatGPT works better with content that's clearly segmented and logically structured. Pages with an average section length of 120 to 180 words between headings perform best, averaging 4.6 citations. Extremely short sections (under 50 words) typically result in only 2.7 citations.

Interestingly, articles with longer paragraphs (over 180 words) perform slightly better (5.7 citations), but this likely correlates more with comprehensive coverage than with readability itself. To make the most of the structure, subheadings should be used to guide the thematic flow and clarify relationships. The content should be divided into paragraphs of 120 to 180 words. A hierarchy should be maintained, with H2 headings for topics, H3 headings for specific aspects, and bullet points for clarity.

How important are FAQ sections and question-based headings?

Many in the industry believe that including FAQ-like sections in ChatGPT content helps users find direct answers and increases the chances of appearing in AI responses. The analysis examined texts for patterns indicating such sections, such as headings like FAQ, Frequently Asked Questions, Questions and Answers, Common Questions, or Popular Questions.

At first glance, pages with FAQs and question-based titles appear to perform worse. FAQ pages achieve an average of 3.8 citations, compared to 4.1 for those without. Question-style headings show 3.4 citations versus 4.3 for simple headings. However, the deeper insight is that, according to SHAP values, the model views the absence of FAQ sections as a negative signal. This means that these formats are not inherently bad, but rather context-dependent.

FAQ sections often appear on smaller or simpler pages, such as support documents or product information pages, which naturally receive fewer citations. An FAQ section alone won't dramatically increase citations. Its true benefit becomes apparent when other factors, such as high-quality content, strong authority, and good structure, are already optimized. In such cases, an FAQ section provides an additional boost to citations.

For smaller domains, question-based titles have almost seven times the impact on citations compared to top-level domains. Furthermore, the presence of FAQ sections within the main content nearly doubles the chances of being cited by ChatGPT.

Why is regularly updating content so important?

Content freshness doesn't mean being new, but rather remaining relevant. The data shows that brand-new content performs only slightly better than older material. Very new content (up to two months old) achieves an average of 3.6 citations, while content aged 1.5 to 5 years performs similarly at 3.1 citations.

However, content updated within the last three months receives almost twice as many citations (6.0 versus 3.6). By refreshing existing articles quarterly with new statistics, examples, or sections, the likelihood of being cited by ChatGPT can almost double.

To leverage this, existing articles should be updated quarterly with new statistics, examples, and insights. Current trends or revised sections should be added rather than completely republishing the content. The most frequently linked and cited pages should be tracked and kept up to date.

B2B support and SaaS for SEO and GEO (AI search) combined: The all-in-one solution for B2B companies - Image: Xpert.Digital

AI search changes everything: How this SaaS solution is revolutionizing your B2B rankings forever.

The digital landscape for B2B companies is undergoing rapid change. Driven by artificial intelligence, the rules of online visibility are being rewritten. It has always been a challenge for companies to not only be visible in the digital masses, but also to be relevant to the right decision-makers. Traditional SEO strategies and local presence management (geomarketing) are complex, time-consuming, and often a battle against constantly changing algorithms and intense competition.

But what if there were a solution that not only simplifies this process, but makes it smarter, more predictive, and far more effective? This is where the combination of specialized B2B support with a powerful SaaS (Software as a Service) platform, specifically designed for the needs of SEO and GEO in the age of AI search, comes into play.

This new generation of tools no longer relies solely on manual keyword analysis and backlink strategies. Instead, it leverages artificial intelligence to more precisely understand search intent, automatically optimize local ranking factors, and conduct real-time competitive analysis. The result is a proactive, data-driven strategy that gives B2B companies a decisive advantage: They are not only found, but perceived as the authoritative authority in their niche and location.

Here's the symbiosis of B2B support and AI-powered SaaS technology that is transforming SEO and GEO marketing and how your company can benefit from it to grow sustainably in the digital space.

More about it here:

• 
  
  
  • 🔵 Partner of Xpert.Digital (Business Development)
  • 🔵 SE Ranking (Website)
  
  
  B2B Support & Blog for SEO, GEO and AIS – Artificial Intelligence Search
• Forget expensive SEO tools – this alternative dominates with unbeatable B2B features

Why shouldn't URLs and titles be excessively keyword-optimized?

The analysis shows that the similarity of a keyword to the page's target topic usually plays no role. The effect was so small that it confirms what many already suspect: Over-optimizing content with keywords does not help AI models understand it better.

For URLs, the analysis reveals a fairly linear pattern: pages with low semantic relevance (0.00 to 0.57) achieve an average of 6.4 citations, medium relevance (0.58 to 0.76) sees 3.4 to 4.5, and the highest relevance (0.84 to 1.00) drops to 2.7. This shows that ChatGPT favors URLs that clearly describe the overall topic, rather than those strictly optimized for a single keyword.

The overall picture is similar for titles. Titles with low semantic relevance (0.00 to 0.59) achieve an average of 5.9 citations, while those with the highest (0.84 to 1.00) reach approximately 2.8. That's more than double the difference. Overall, focusing on clarity and relevance rather than inserting keywords makes the content easier for AI to understand and more trustworthy.

Passdemone:

• Who are SE Ranking's competitors, and why does SE Ranking have an advantage, especially in the B2B sector? – Xpert Recommendation on SEO/GEO

What role do Core Web Vitals play in ChatGPT citations?

Loading speed does play a role, but with some nuances. Metrics like First Contentful Paint (FCP) and Largest Contentful Paint (LCP) show an inverse correlation with citations: the faster, the better, but only up to a certain point.

The fastest websites (FCP under 0.4 seconds) achieve an average of 6.7 citations. The slowest (over 1.1 seconds) drop to 2.1 citations. However, the middle range (approximately 0.65 to 0.82 seconds) remains stable at 4.2 citations. This means that one doesn't need to strive for extreme speeds, but rather avoid slowness, which indicates poor technical quality.

The Speed ​​Index confirms this: websites with index values ​​below 1.14 seconds perform reliably well, but those above 2.2 seconds experience a steep decline. Interestingly, pages with the best INP values ​​(below 0.4 seconds) tend to receive fewer citations (an average of 1.6), while those with moderate INP values ​​(0.8 to 1.0 seconds) receive more citations (4.5). This suggests that pages that are too simple or static, even if they function perfectly, may not be considered authoritative or engaging sources by ChatGPT.

What influence does a presence on review platforms have?

This analysis considered five major review platforms: Trustpilot, G2, Capterra, Sitejabber, and Yelp. The trend is the same as with Reddit and Quora: Domains present on review platforms consistently outperform those without such visibility.

Domains present on multiple review platforms achieve 4.6 to 6.3 citations, compared to 1.8 for those not present on such platforms. ChatGPT thus observes social validation. While these platforms are not direct SEO factors, they act as trust multipliers.

Profiles on major review sites should be claimed and verified. Authentic reviews from real users should be encouraged. Feedback should be monitored and responded to in order to build trust.

Why shouldn't one rely solely on LLMs.txt or FAQ schema markup?

Many so-called AI optimization strategies fail to deliver the expected results. A notable example is LLMs.txt, a proposed file format intended to help AI models understand and cite website content. Contrary to its promise, including it actually decreased model accuracy during analysis. Removing it improved prediction results, suggesting that ChatGPT is not currently relying on it at all.

At the same time, the use of FAQ schema markup, often touted as an essential element for LLM optimization, has shown surprisingly weak results. In fact, pages with FAQ schema markup have 3.6 citations, compared to 4.2 without.

The key takeaway here is that structured data is a nice-to-have, not a game-changer. LLMs seem to care more about whether the information is structured (via headings) than whether it's technically marked up. Therefore, the focus should initially be on content organization; schema markup is the icing on the cake.

What is the research methodology of the study?

This research investigates the factors that influence how LLMs, particularly ChatGPTs, cite websites as sources when generating answers. To this end, a large dataset of 129,000 unique domains, comprising 216,524 pages and covering 20 different niches, was analyzed to ensure a diverse and representative sample.

For each domain, data was collected on a variety of factors, including: domain authority and trust, brand visibility and social presence, content quality and semantic relevance, technical performance, SEO visibility, and traffic metrics.

To analyze the relationships between these factors and citation probability, an XGBoost regression model was used. The target variable in the model was the number of citations a domain receives in ChatGPT responses, based on a parsed dataset of 100,000 prompts. This regression approach made it possible to identify which features most strongly predict citation frequency.

To interpret the model and understand how each factor influences the probability of a citation, SHAP analysis (SHapley Additive Explanations) was applied, a method from game theory that quantifies the contribution of each feature to the model's predictions. The report focuses on the 20 most important factors, ranked according to their significance in the model.

What are the most important factors that influence ChatGPT citations?

In investigating what prompts ChatGPT to cite a website, several factors were found that influence the likelihood of citations. Some are well-known from SEO, while others are unique to AI systems.

At the top of the list is the number of referring domains. The more diverse websites link to the content, the more likely ChatGPT is to consider it credible. Domain traffic comes close behind: high visitor numbers signal authority, credibility, and broad relevance.

But authority isn't just about links and visitors. Page and domain trust scores play a critical role in demonstrating how reliable and reputable a website appears. Technical performance also makes a difference. Pages that load quickly (as measured by INP, FCP, and LCP) are more likely to attract attention.

Content structure and depth are equally important. Longer articles, FAQ or Q&A sections, and question-based titles and headings all correlate with a higher likelihood of being cited. And while brand-new content isn't always preferred, keeping content fresh and regularly updated helps maintain relevance over time.

Even social signals are important. Websites that are actively mentioned or discussed on platforms like Quora and Reddit have a higher chance of being cited on ChatGPT.

Taken together, these factors show that ChatGPT citations are based on a combination of technical optimization, authoritative content, social proof, and user-friendly experience.

What conclusions can be drawn from the study?

The study shows that flashy AI hacks like LLMs.txt have little effect. What truly drives ChatGPT citations are the fundamentals: strong backlinks, high domain and page trust, solid web traffic, and content that is in-depth, clear, and easy to read. Fast, responsive pages also help, and activity on Quora, Reddit, and major review sites gives visibility a noticeable boost. Furthermore, regularly updating content ensures that ChatGPT keeps coming back.

The focus should be on building genuine authority and providing added value. Small websites can score above their weight class with thorough, well-structured content and a strong social media presence, while large websites can maintain their lead by focusing on trust, backlinks, and consistent content quality.

How do the requirements for small and large websites differ?

For smaller, less established websites, engagement on Quora and Reddit offers a way to build authority and gain ChatGPT trust, similar to how larger domains achieve this through backlinks and high traffic. Content length has approximately 65 percent more impact on ChatGPT citations for smaller domains than for top-level domains.

For smaller domains, question-based titles have almost seven times the impact on citations compared to top-level domains. This means that smaller websites should structure their content particularly carefully using questions to maximize their chances.

Larger websites, on the other hand, benefit most from their existing backlink profile and traffic. For them, maintaining the quality and freshness of their content and further expanding their authority is more important than focusing on specific formatting techniques.

What practical steps should be taken to improve ChatGPT visibility?

To improve the chances of being cited by AI systems like ChatGPT, several strategies should be pursued simultaneously. First, website authority should be strengthened, with backlinks remaining the strongest signal of trust and credibility.

Overall visibility on Google should be increased by optimizing content for organic performance, investing in technical SEO, link building, and high-quality writing. Google rankings can be treated as an indicator of LLM visibility.

Comprehensive, in-depth content of at least 1,900 words should be produced for most topics. If necessary, the word count should be increased to at least 2,900 words. Related concepts, synonyms, and examples should be covered to enhance semantic diversity. Supporting data such as expert citations and statistics should be included, not as gimmicks, but as evidence of depth and professionalism.

The content should be structured for readability and clarity, organized into sections of 120 to 180 words between headings. FAQ sections at the end of key articles, as well as question-based subheadings seamlessly integrated into the text, should be included.

The content should be kept fresh with regular updates, with existing articles being updated quarterly with new statistics, examples, and insights. Profiles should be claimed and verified on major review sites, and authentic reviews from real users should be encouraged.

Core Web Vitals should be optimized, paying attention to fast loading times and avoiding extremely slow ones. Brand visibility on Quora and Reddit should be built by participating in relevant discussions, encouraging organic brand mentions through helpful contributions, and using these platforms to demonstrate expertise and authority.

Which factors have no significant influence on ChatGPT citations?

The study also identifies factors that, contrary to popular belief, do not have a significant impact on ChatGPT citations. LLMs.txt proved to be negligible, and the analysis even suggests that removing it improved prediction accuracy, so it should not be a focus of AI visibility.

FAQ schema markup alone does not significantly increase the likelihood of ChatGPT citations. The data shows that pages with FAQ schema achieve an average of 3.6 citations, while those without achieve 4.2. Structured data is therefore a nice-to-have, not a game-changer.

Linking to other highly authoritative websites from your own website has minimal effect. Whether you link to domains with a Trust score of 70 or 100, the influence is almost zero compared to incoming backlinks. Trusted domain zones like .gov or .edu do not automatically guarantee higher citations.

Over-optimized URLs and titles with high keyword density perform worse than those that clearly and comprehensively describe the overall topic. Focusing on clarity and relevance rather than simply inserting keywords makes the content easier for AI to understand and more trustworthy.

What significance does the study have for the future of search engine optimization?

The findings of this comprehensive study suggest that optimizing for AI systems like ChatGPT doesn't require a fundamentally new discipline. Rather, it confirms that proven SEO fundamentals remain crucial. Websites that have already invested in traditional search engine optimization are well-positioned to succeed in the era of AI-powered search.

At the same time, the study opens up new perspectives for smaller websites. Engagement on social discussion platforms like Reddit and Quora offers an alternative path to building authority that doesn't depend solely on large backlink profiles. This democratizes access to AI citations, in a sense, and gives less established players a realistic chance at visibility.

The results also underscore the importance of a holistic approach. No single tactic can guarantee success. Instead, a balanced strategy is needed that combines technical optimization, high-quality content, building authority, and social media presence. Websites that invest in all these areas will achieve the best results in the long run.

How should the optimization measures be prioritized?

When prioritizing optimization measures, the focus should initially be on the factors with the strongest impact. Building referring domains and increasing domain trust should be top priorities, as these represent the strongest signals for ChatGPT citations.

In parallel, the creation and regular updating of in-depth, comprehensive content should be promoted. Content with at least 1,900 to 2,900 words, containing expert quotes and statistics, has the best chance of being cited.

Technical optimization, especially of the Core Web Vitals, should be considered a fundamental requirement. While fast loading times are not a unique selling point, slow pages are clearly at a disadvantage.

Finally, building a presence on discussion and review platforms should be viewed as a long-term investment. This measure is particularly valuable for smaller websites that do not yet have a strong backlink profile.

The most important lesson remains: there are no shortcuts. Sustainable visibility in AI systems requires continuous investment in quality, authority, and user-friendliness.

☑️ Our business language is English or German

☑️ NEW: Correspondence in your national language!

Konrad Wolfenstein

I would be happy to serve you and my team as a personal advisor.

You can contact me by filling out the contact form or simply call me on +49 89 89 674 804 (Munich) . My email address is: wolfenstein ∂ xpert.digital

I'm looking forward to our joint project.

☑️ SME support in strategy, consulting, planning and implementation

☑️ Creation or realignment of the digital strategy and digitalization

☑️ Expansion and optimization of international sales processes

☑️ Global & Digital B2B trading platforms

☑️ Pioneer Business Development / Marketing / PR / Trade Fairs

Our global industry and business expertise in business development, sales and marketing - Image: Xpert.Digital

Industry focus: B2B, digitalization (from AI to XR), mechanical engineering, logistics, renewable energies and industry

More about it here:

• Xpert Business Hub

A topic hub with insights and expertise:

• Knowledge platform on the global and regional economy, innovation and industry-specific trends
• Collection of analyses, impulses and background information from our focus areas
• A place for expertise and information on current developments in business and technology
• Topic hub for companies that want to learn about markets, digitalization and industry innovations

Benefit from Xpert.Digital's extensive, fivefold expertise in a comprehensive service package | R&D, XR, PR & Digital Visibility Optimization - Image: Xpert.Digital

Xpert.Digital has in-depth knowledge of various industries. This allows us to develop tailor-made strategies that are tailored precisely to the requirements and challenges of your specific market segment. By continually analyzing market trends and following industry developments, we can act with foresight and offer innovative solutions. Through the combination of experience and knowledge, we generate added value and give our customers a decisive competitive advantage.

More about it here:

• Use the 5x expertise of Xpert.Digital in one package - starting at just €500/month