GEO Optimization: 11 Tips Fact Checked for Online Stores

GEO optimization has become part of online marketing, and advice is plentiful: schema markup, llms.txt, and publishing your own best lists are all recommended. Yet these recommendations rarely explain whether a measure has actually been shown to work.

We therefore compared 11 common GEO tips with current studies, large data analyses, and official statements from Google and OpenAI. The results are more restrained than many guides suggest. Online stores also face a particular challenge: AI answers reach their customers at a different stage than conventional search results.

What is GEO optimization?

GEO optimization, or Generative Engine Optimization, covers all measures used to prepare content so that AI systems such as ChatGPT, Perplexity, Gemini, and Google’s AI Overviews can find it, understand it, and cite it in their answers. The same discipline is also called LLMO, or Large Language Model Optimization, and AEO, or Answer Engine Optimization. To understand which measures work, it helps to look at how AI systems select their sources.

How AI search systems select sources

AI search systems do not answer new or complex questions solely from their training knowledge. They research the live web. This generally involves three steps:

1. Break down the question: The system turns a user question into several search queries. Google calls this query fan-out: AI Mode breaks a question into subtopics and runs numerous searches simultaneously.¹
2. Retrieve sources: For each subquery, the system retrieves results from a search index.
3. Select passages: It incorporates individual, particularly relevant passages from the retrieved pages into its answer.

Systems retrieve far more than they ultimately cite. An analysis of 15,000 prompts shows that ChatGPT does not cite 85 percent of the pages it retrieves.² For your store, this means being found is not enough. Your page must provide the most relevant answer to a specific subquery.

GEO vs. SEO: What changes and what stays the same

GEO builds on SEO rather than replacing it. In its guide to generative AI search, Google describes optimization for generative AI as “still SEO” from the perspective of Google Search.³ The search index remains the foundation from which AI systems obtain their sources.

What changes is the unit of competition. In SEO, entire pages compete for positions. In GEO, individual passages compete for a place in an answer. Good rankings for the main keyword therefore do not guarantee a citation: only 38 percent of pages cited in AI Overviews rank in the top 10 for the corresponding search query.⁴

• SEO: The goal is a click from the results list. Rankings, impressions, and clicks are the metrics.
• GEO: The goal is a mention or citation in an AI answer. The metric is the share of answers in which your brand or page appears.
• Shared foundation: Indexable, relevant, trustworthy content forms the basis of both SEO and GEO optimization.

This differs from using AI tools to speed up your own SEO work. Read more in our article on AI SEO.

Why GEO works differently for online stores

For online stores, GEO mainly influences research questions and has little impact on direct shopping searches. To understand where the effort pays off, you need to distinguish between these situations.

Shopping searches remain conventional searches

For transactional searches, where users want to buy a specific product, Google rarely displays AI Overviews in Germany. SISTRIX analyzed around 2,500 of Germany’s highest revenue online stores: AI Overviews appear in just 0.2 percent of transactional searches.⁵ Organic results and Shopping ads continue to dominate.

User behavior is also shifting more slowly than many forecasts suggest. In a Bitkom survey, 16 percent of respondents said they frequently used an AI chatbot instead of an internet search when looking for products.⁶ That is relevant, but still far from a majority.

Where AI answers reach store customers

AI answers mainly reach your customers during the research phase, with questions such as “Which robot vacuum is suitable for pet hair?” or “What should beginners look for in running shoes?” This is where brands and stores make it onto the shortlist.

So far, only a few stores benefit. According to the same SISTRIX analysis, the share of citations attributed to stores themselves rose from just 2.3 to 5.3 percent within a year. The most visible fifth of stores accounts for around 72 percent of all AI citations.⁵

For product questions that do not specify a brand, AI systems primarily cite independent sources such as reviews and specialist media. In a consumer electronics study, 92.1 percent of AI citations in the US came from earned media, meaning content published about a brand by third parties.⁷

Product data offers another route into AI answers. Google uses Merchant Center data in shopping answers, while OpenAI accepts product catalogs as structured data through the Agentic Commerce Protocol, so far with a focus on the US market.⁸ Our article on AI product search explains how to prepare your product data.

Why AI visibility matters even without a direct click

An AI answer can already influence which providers a user considers. If your brand is named in response to a relevant research question, this creates a potential point of contact with your offer. If it is absent, that contact does not take place in that particular answer.

A Similarweb study using US desktop data shows that visits can follow a recommendation later. After a recommendation for Sephora, 7.9 percent of the users studied visited its website within seven days, compared with 3.3 percent for competitor Ulta. Across the brands studied, 55.9 percent of subsequent visits arrived through search engines. Direct AI clicks alone therefore do not capture the full potential value of an AI recommendation.⁴²

What matters, however, is how your brand appears. An explicit mention in the answer text differs from a source link behind an expandable reference. In a Semrush study, 61.7 percent of the 3,981 domain appearances recorded were source citations without an explicit brand mention in the answer text.⁴³

For your store: Alongside clicks, track which customer questions lead to your brand being named, recommended, or displayed as a source. Consider the context: a relevant mention in a buying guide can matter more for your brand presence than numerous source links for general information queries.

The prerequisite: Allow AI crawlers access

AI crawlers are bots that providers such as OpenAI, Anthropic, Perplexity, and Google use to retrieve websites for AI answers or model training. If you block search crawlers, your store cannot appear as a source in the corresponding AI answers, regardless of content quality. This prerequisite therefore comes before all 11 tips.

Which AI crawlers matter for visibility?

Providers now distinguish between crawlers for search and crawlers for training. For visibility in AI answers, search crawlers are particularly relevant:

• ChatGPT: OAI-SearchBot brings websites into ChatGPT search results. Pages that exclude it do not appear in search answers. GPTBot, by contrast, collects content for model training.⁹
• Claude: Claude-SearchBot and Claude-User retrieve content for user searches, while ClaudeBot collects training data.¹⁰
• Perplexity: PerplexityBot lists websites in Perplexity search results and, according to the provider, is not used for model training.¹¹
• Google AI Overviews and AI Mode: Your page must be indexed by Google and eligible to appear with a snippet. There are no additional special requirements.¹²
• Google-Extended: This token only controls whether Google may use your content for Gemini. According to Google, it has no effect on inclusion in Google Search.¹³

This gives stores a clear distinction: allow search crawlers and make a separate decision about training crawlers. Blocking GPTBot or ClaudeBot does not automatically remove your visibility in search answers. The providers’ official documentation supports this distinction.

How to check access to your store

Checking robots.txt alone is often insufficient. Security settings in a CDN or firewall frequently block AI crawlers without anyone noticing. In July 2025, for example, Cloudflare announced that it would block AI crawlers by default unless they paid for content.¹⁴

1. Check robots.txt: Open the file in your store’s root directory, for example your-store.com/robots.txt, and look for Disallow rules targeting OAI-SearchBot, Claude-SearchBot, PerplexityBot, or all bots.
2. Check your CDN and firewall: Review bot protection and AI crawler settings in services such as Cloudflare. Depending on the configuration, AI bot blocking features or strict bot protection may also exclude search crawlers.
3. Analyze server logs: Check whether crawlers retrieve your pages and receive status code 200. Frequent 403 or 429 responses indicate a block.
4. Allow time for changes: According to OpenAI, ChatGPT search may take around 24 hours to reflect a robots.txt change.⁹

How the 11 GEO tips were assessed

The following fact check examines whether published data supports the effectiveness of 11 common GEO tips. It draws on studies, large analyses by SEO tool providers, and official statements from Google and OpenAI, as of October 2026. Each tip receives one of three ratings:

• Supported: Studies with a large dataset or official provider statements support the recommendation.
• Partially supported: Data supports part of the recommendation, or relevant counterevidence exists.
• Unsupported: Practical observations exist, but no published data.

The results at a glance:

1. Use query fan-out, with SEO as the foundation: Supported
2. Build your website as a data interface: Supported
3. Update content regularly: Supported
4. Establish your brand across three layers: Partially supported
5. Quantify brand claims: Partially supported
6. Place relevant facts on every relevant page: Partially supported
7. Start with what sets you apart: Partially supported
8. Use a consistent format for best lists: Partially supported
9. Publish only credible comparisons: Partially supported
10. Do not publish exclusion criteria: Unsupported
11. Create a separate page for each subtopic: Unsupported

Keep the context in mind: most studies measure correlations, not causes. AI systems also change their behavior with each model update, so every assessment is a snapshot.

Supported: These 3 GEO tips have evidence behind them

Large data analyses or official statements support three GEO tips: rankings for subqueries, a clearly readable website, and regularly updated content. These measures should be your starting point.

Tip 1: Use query fan-out, with SEO as the foundation

The recommendation: Base your subheadings on the subqueries into which AI systems break down a user question, and secure good rankings for those specific queries.

Assessment: Supported. An analysis of around 174,000 URLs shows that pages also ranking for fan-out queries have a 161 percent higher probability of being cited in AI Overviews.¹⁵

Heading wording also matters. ChatGPT cited pages whose headings closely matched the user question in 41 percent of cases, compared with 29 percent for a weak match.¹⁶

Limitation: The common assumption that ChatGPT uses Bing’s index no longer applies in 2026. ChatGPT now works with its own index, only 1.5 percent of whose results appear in Bing’s top 20.¹⁷

Google also explicitly advises against artificially breaking text into small chunks for AI systems.³ Good structure therefore means clear headings and direct answers, rather than fragmented content.

For your store: Collect questions customers ask before buying, for example from Search Console, customer service inquiries, and Google’s “People also ask” section. Use their wording as subheadings in buying guides and category content, and answer each question in the first sentence of its section.

Practical example: Customer questions at Schuh Keller and Bergfreunde

Schuh Keller shows what a specific answer can look like: its advice page uses the question “How much space should there be at the front of a hiking boot?” as a separate subheading. The answer immediately below gives a thumb’s width as a guide, explains what happens when walking downhill, and recommends trying boots on in the afternoon.³⁹

In our Google search for the German query “wie viel platz vorne im wanderschuh” on October 19, 2026, this page appeared as a source in the AI Overview. The same answer also cited Bergfreunde’s guide to breaking in hiking boots, which covers suitable boot length and how securely the foot should sit.⁴⁰

How to apply this: Answer specific questions about choosing and using your products in clearly labelled sections. For a question about fit, a direct answer and a short explanation are more useful than a general description of your range.

Google AI Overview for the German query ‘wie viel platz vorne im wanderschuh’: Schuh Keller in the open source panel and Bergfreunde in the source list on the right. Screenshot taken on October 19, 2026, Germany, desktop.

Tip 2: Build your website as a data interface

The recommendation: Display facts such as prices, dimensions, and delivery terms clearly and visibly on your regular pages instead of creating separate pages solely for machines.

Assessment: Supported. Google makes clear that you do not need new machine readable files, AI text files, or Markdown versions to appear in AI search. According to Google, structured data is not a prerequisite either.³

Independent data confirms this. A study of 1,885 pages receiving new schema markup found no significant increase in AI citations compared with 4,000 control pages. When retrieving pages directly, AI systems evaluated only the visible HTML content.¹⁸

The picture is similar for llms.txt, a text file with guidance specifically for language models. An analysis of 300,000 domains found no relationship between this file and AI citations.¹⁹

Limitation: Structured data remains useful for rich results in conventional search. Stores also have a structured channel that does matter: product feeds, for example through Google Merchant Center.

For your store: Check that price, availability, shipping costs, and key product features appear as HTML text rather than only in images, scripts, or PDF datasheets. Keep your product feeds complete and current.

Practical example: Visible lawn care advice at Rasenwelt

Rasenwelt brings together the recommended timing, the appropriate fertilizer type, and the individual steps on its page about autumn fertilization. A short summary and a numbered guide appear directly in the page text. Readers can learn when to fertilize, why potassium matters, and what to do before and after application.⁴¹

In our Google search for the German query “rasen im herbst düngen” on October 19, 2026, this exact page appeared in the AI Overview’s source list.

How to apply this: Put information customers need for their decisions, such as timing, application, and care, visibly on the relevant page. An additional file can be helpful, but the key information should also be readable without a download.

Google AI Overview for the German query ‘rasen im herbst düngen’, showing Rasenwelt in the source list on the right. Screenshot taken on October 19, 2026, Germany, desktop.

How to interpret these examples: The screenshots document individual search results. They show that the pages were selected as sources. They do not establish whether the content structure described caused that selection. Answers and sources may change when a query is repeated.

Tip 3: Update content regularly

The recommendation: Maintain important content regularly so AI systems recognize it as current.

Assessment: Supported. An analysis of 17 million citations shows that AI assistants such as ChatGPT and Perplexity cite content that is, on average, 25.7 percent fresher than conventional search results.²⁰

Updates matter more than the publication date. In a study of over 47,000 citations, 75 percent of cited pages had been updated within the previous year, while only 42 percent had been published during that period.²¹

Limitation: Google AI Overviews do not show this effect. On average, they even cite slightly older content than organic search.²⁰ Relevance remains more important than recency. Changing a date alone does not replace improving the content.

For your store: Set a regular review schedule for your most important buying guides and category content, for example every six months. Update figures, models, and prices, and display the modification date.

Partially supported: 6 GEO tips with limitations

Six GEO tips are partially supported: the data backs a core part of the recommendation but also reveals limits or conflicting results. Examine these closely before allocating resources.

Tip 4: Establish your brand across three layers

The recommendation: Define your brand factually on your own website, have independent sources confirm these statements, and monitor what communities say about you.

Assessment: Partially supported. An analysis of 75,000 brands shows that web mentions of a brand correlate much more strongly with its visibility in AI Overviews than backlinks do. The correlation is 0.664 for mentions and just 0.218 for backlinks.²²

The three layer model itself comes from a study of over 231,000 AI answers. People with all three signals, meaning their own definition, third party confirmation, and community discussion, received around 12,174 AI mentions, compared with around 1,565 without the full set of signals.²³

Limitation: This study examined Olympic athletes, not brands. The community layer is also unstable: in 2025, the share of ChatGPT answers citing Reddit fell from almost 60 to around 10 percent within a few weeks.²⁴

For your store: Focus on independent reviews, specialist media, and comparison sites in your industry, since AI systems primarily cite these sources for product questions.⁷ Customer reviews and social proof further strengthen the picture.

Tip 5: Quantify brand claims

The recommendation: Describe your brand with verifiable figures such as customer numbers, product range size, or delivery time instead of general promises.

Assessment: Partially supported. The foundational GEO paper by researchers from Princeton and Delhi shows in a laboratory experiment that content containing statistics, quotations, and source references achieves around 30 to 40 percent more visibility in AI answers. Keyword stuffing, by contrast, performed worse than the original text.²⁵

Limitation: An independent benchmark reaches a different conclusion: most optimization methods tested were largely ineffective and sometimes even worsened rankings.²⁶ No study has directly examined whether brand metrics repeated across many websites strengthen the association with a brand.

For your store: Replace phrases such as “wide selection” or “fast shipping” with specific information, such as “over 4,000 products” or “dispatch within 24 hours.” This also helps visitors make decisions.

Tip 6: Place relevant facts on every relevant page

The recommendation: Put relevant figures and facts on the specific page matching the query, rather than somewhere else on your domain.

Assessment: Partially supported. Strong evidence shows that the individual page matters: ChatGPT retrieves many pages per query but cites only a small share.² The decisive factor is whether the retrieved page directly answers the subquery.

There is also evidence that facts must be visible in the text. When retrieving pages directly, AI systems evaluated only visible HTML content and ignored embedded background data.¹⁸

Limitation: The common assumption that AI systems retrieve only a few pages is incorrect. Whether additional statistics alone increase the chance of citation is disputed, as Tip 5 shows.

For your store: Add the figures customers need to make decisions to category pages and buying guides, such as size charts, material specifications, care instructions, or comparison values.

Tip 7: Start with what sets you apart

The recommendation: Begin each page with what distinguishes you. Your own data, experience, or unique selling proposition belongs near the top.

Assessment: Partially supported. An analysis of 1.2 million ChatGPT answers shows that 44.2 percent of all citations come from the first 30 percent of a text.²⁷ The beginning of a page therefore carries disproportionate weight.

Limitation: More than half of citations nevertheless come from later sections. The study also examined only ChatGPT, with no separate data for store pages.

For your store: Put the most important buying arguments at the beginning of category pages and guides, for example your advisory expertise, special warranties, or test results. Generic introductions that could appear on any other page belong further down or can be removed.

Tip 8: Use a consistent format for best lists

The recommendation: If you publish comparisons or best lists, describe all providers using the same format instead of covering yourself extensively and others only briefly.

Assessment: Partially supported. An analysis of 100 software queries in Google AI Overviews shows that when a self promotional best list was cited as a source, its publisher’s brand was absent from the actual recommendation in 69 percent of cases.²⁸ Instead, established competitors from the list were mentioned.

Limitation: A different analysis reaches the opposite conclusion for ChatGPT. Best lists accounted for 43.8 percent of all cited page types, with no apparent disadvantage for self promotion.²⁹ There is no data supporting the recommendation to use a consistent format.

For your store: This tip is usually a secondary priority for online stores. For product questions in the same ChatGPT analysis, product pages accounted for 87.2 percent of citations of owned pages, while blog lists accounted for just 4 percent.²⁹ Invest in your product and category pages first.

Tip 9: Publish only credible comparisons

The recommendation: Publish only comparisons that are factually accurate, both in the selection of providers and in their descriptions.

Assessment: Partially supported. From January 2026, websites with many lists ranking themselves lost between 29 and 49 percent of their Google visibility.³⁰ The relationship has been observed but not confirmed.

Limitation: Neither Google nor OpenAI has officially stated that such lists are devalued. The legal position in Germany, by contrast, is clear: under Section 6 of the German Act Against Unfair Competition, comparative advertising must be objective and refer to material, relevant, verifiable, and representative features or price. Competitors must not be disparaged.³¹

For your store: Compare products or brands only using verifiable features such as price, dimensions, materials, or test results. If in doubt, have comparison pages legally reviewed before publishing them.

Unsupported: 2 GEO tips without a data basis

Two common GEO tips lack evidence: practical observations exist, but no published data. This does not mean they are wrong, but you should prioritize them cautiously.

Tip 10: Do not publish exclusion criteria

The recommendation: Do not state on your website who your offer is unsuitable for, because AI systems could use those statements to exclude you from recommendations.

Assessment: Unsupported. There is no study on phrases such as “not suitable for.” The only indirect indication comes from research on product recommendations by language models: products explicitly targeting an exclusive customer group were recommended 45 percent less often.³²

Limitation: The study tested exclusive advertising language, not exclusion wording. If a query closely matches your target audience, a clear distinction might even help. This has not been investigated either.

For your store: State positively what a product is suitable for, for example “for hard floors and low pile carpets” instead of “not for deep pile carpets.” This is also the stronger sales wording.

Tip 11: Create a separate page for each subtopic

The recommendation: Create a separate page for every topic for which you want to be cited instead of covering it in a section of a general guide.

Assessment: Unsupported. The underlying mechanism has partial support: cited pages have titles that are considerably more similar to the user question than the titles of pages that are not cited.³³

Furthermore, 82.5 percent of citations in AI Overviews go to deeper pages, and less than 0.5 percent to homepages.³⁴ This does not establish that a separate page outperforms a good section in a comprehensive guide.

Limitation: No study compares the two approaches directly. Google also explicitly advises against artificially fragmenting content for AI systems.³

For your store: Create separate pages only for topics with a distinct search intent, such as a buying guide for each product category. Answer smaller questions in clearly headed sections of existing pages. Our article on content marketing explains how to plan this content.

Measuring AI visibility: What is meaningful?

AI visibility describes how often your brand or pages are mentioned or cited in AI answers. Because these answers vary considerably, individual queries tell you little.

Use prompt tracking correctly

Prompt tracking means regularly submitting a fixed list of typical customer questions to AI systems and measuring how often your brand appears. The variation is substantial: the chance of receiving the same brand list twice for the same question is less than 1 in 100.³⁵

Cited sources also change rapidly. A longitudinal analysis spanning 17 weeks and six countries, including Germany, shows that Google replaces 56 percent of cited sources each week, and ChatGPT 74 percent.³⁶

• Meaningful metric: The share of answers mentioning your brand, measured across dozens to hundreds of prompts with repeated queries.
• Less meaningful: A ranking position in a single AI answer.
• Additional benefit: Prompt tracking reveals whether AI systems are spreading incorrect prices or outdated product information about your store.

Identify AI traffic in web analytics

AI traffic consists of visits arriving at your website through links in AI answers. In web analytics tools such as Google Analytics 4, you can identify it by referrers such as chatgpt.com, perplexity.ai, or gemini.google.com.

Its share is still small. An analysis of over 100,000 websites puts AI traffic at 0.32 percent of total traffic.³⁷

Whether these visitors convert better is disputed. A study of 973 online stores found that conversion rates for ChatGPT traffic were lower than those of every conventional channel except paid social.³⁸

For stores, this means AI visibility is a growing channel, but it does not replace optimizing the visits you already receive. Our article on traffic vs. conversions explains why visitor numbers alone tell you little.

GEO optimization for stores: The right order

GEO optimization for an online store starts with technical accessibility and ends with measurement. The fact check points to the following sequence:

1. Allow AI crawlers: Check robots.txt, CDN settings, and server logs as described above. Search crawlers such as OAI-SearchBot, Claude-SearchBot, and PerplexityBot are crucial.
2. Keep product data complete: Maintain product feeds and display price, availability, and key features as visible text on product and category pages.
3. Answer research questions: Create buying guides and category content whose subheadings reflect real customer questions, with the answer in the first sentence.
4. Update content: Review your most important pages on a fixed schedule and make updates visible.
5. Build independent mentions: Focus on reviews, specialist media, and customer ratings rather than creating your own best lists.
6. Measure visibility: Track your share of AI answers across many prompts and monitor AI referrers in your web analytics.

Visitors arriving from AI answers are usually well informed, but not necessarily ready to buy. This makes what happens after the click particularly important. This is where uptain comes in: software for data driven ecommerce marketing and reducing cart abandonment, used by over 3,000 online stores. Exit intent popups let you reach visitors before they leave your store, while automated emails help recover abandoned carts. Find further approaches in our tips to reduce checkout abandonment.

Frequently asked questions about GEO optimization

What is GEO optimization?

GEO optimization, or Generative Engine Optimization, covers all measures used to prepare content so that AI systems such as ChatGPT, Perplexity, or Google AI Overviews can find, understand, and cite it. The goal is a mention or citation in an AI answer rather than a click from a results list.

How do you optimize for GEO?

Start with the basics: allow AI crawlers, put facts on the relevant pages as visible text, and use real customer questions as subheadings. Update important content regularly and build independent mentions. According to current data, special AI files such as llms.txt or additional schema markup provide no measurable advantage.

What is the difference between GEO and SEO?

SEO optimizes entire pages for rankings and clicks in search results. GEO optimizes individual passages and perceptions of your brand so AI systems cite or recommend you in answers. Both share the same foundation: AI systems research using search indexes, so good rankings remain a prerequisite for successful GEO.

Is SEO still worthwhile?

Yes. AI systems obtain sources from search indexes, and Google explicitly describes optimization for generative AI search as SEO. This is particularly relevant for online stores: AI Overviews appear in just 0.2 percent of transactional searches in Germany, so shopping searches still largely use conventional search.

Conclusion

GEO optimization extends SEO. Once AI search crawlers can access your store, the main factors that work are good rankings for subqueries, clearly readable facts, and current content. Many popular tips, by contrast, are only partially supported or unsupported, and special AI files currently offer no benefit. For online stores, GEO mainly pays off during the research phase, while shopping searches remain conventional for now. Implementing the basics properly and measuring results across many prompts directs investment toward the measures for which today’s data shows an effect.

References

¹ Google: AI Mode in Google Search: Updates from Google I/O 2025 (2025), https://blog.google/products-and-platforms/products/search/google-search-ai-mode-update/ (accessed: 05.10.2026)

² AirOps: Influence of Retrieval, Fan-out and Google SERPs on ChatGPT Citations (2026), https://www.airops.com/report/influence-of-retrieval-fanout-and-google-serps-in-chatgpt (accessed: 05.10.2026)

³ Google Search Central: Optimizing for generative AI features in Google Search (2026), https://developers.google.com/search/docs/fundamentals/ai-optimization-guide (accessed: 05.10.2026)

⁴ Ahrefs: Update: 38% of AI Overview Citations Pull From The Top 10 (2026), https://ahrefs.com/blog/ai-overview-citations-top-10 (accessed: 05.10.2026)

⁵ SISTRIX: KI und SEO im E-Commerce (2026), https://www.sistrix.de/news/ki-seo-e-commerce-zitierfaehigkeit/ (accessed: 05.10.2026)

⁶ Bitkom: Digitaler Handel in Deutschland 2026 (2026), https://www.bitkom.org/sites/main/files/2026-07/bitkom-studienbericht-digitaler-handel-in-deutschland.pdf (accessed: 05.10.2026)

⁷ Chen et al.: Generative Engine Optimization: How to Dominate AI Search (2025), https://arxiv.org/abs/2509.08919 (accessed: 05.10.2026)

⁸ OpenAI: Agentic Commerce, developer documentation (2026), https://developers.openai.com/commerce (accessed: 05.10.2026)

⁹ OpenAI: Overview of OpenAI Crawlers (2026), https://developers.openai.com/api/docs/bots (accessed: 13.10.2026)

¹⁰ Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler? (2026), https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (accessed: 13.10.2026)

¹¹ Perplexity: Perplexity Crawlers (2026), https://docs.perplexity.ai/guides/bots (accessed: 13.10.2026)

¹² Google Search Central: AI features and your website (2025), https://developers.google.com/search/docs/appearance/ai-features (accessed: 05.10.2026)

¹³ Google Search Central: Google’s common crawlers (2026), https://developers.google.com/search/docs/crawling-indexing/google-common-crawlers (accessed: 13.10.2026)

¹⁴ Cloudflare: Content Independence Day: No AI crawl without compensation! (2025), https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/ (accessed: 13.10.2026)

¹⁵ Surfer: Ranking for Multiple Fan-Out Queries and AI Overview Citations (2026), https://surferseo.com/blog/query-fan-out-impact (accessed: 05.10.2026)

¹⁶ AirOps: The Fan-Out Effect: What Happens Between a Query and a Citation (2026), https://www.airops.com/report/the-fan-out-effect-what-happens-between-a-query-and-a-citation (accessed: 05.10.2026)

¹⁷ RESONEO: ChatGPT Retrieval Study (2026), https://think.resoneo.com/chatgpt-retrieval/ (accessed: 05.10.2026)

¹⁸ Ahrefs: Study on schema markup and AI citations (2026), https://ahrefs.com/blog/schema-ai-citations/ (accessed: 05.10.2026)

¹⁹ SE Ranking: Study on llms.txt and AI citations (2025), https://seranking.com/blog/llms-txt/ (accessed: 05.10.2026)

²⁰ Ahrefs: AI Assistants Prefer to Cite „Fresher“ Content (2025), https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content (accessed: 05.10.2026)

²¹ Seer Interactive: Study: Content Recency’s Impact on AI Visibility in 2026 (2026), https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026 (accessed: 05.10.2026)

²² Ahrefs: An Analysis of AI Overview Brand Visibility Factors, 75K Brands Studied (2025), https://ahrefs.com/blog/ai-overview-brand-correlation/ (accessed: 05.10.2026)

²³ Seer Interactive: The GEO Olympics Study (2026), https://www.seerinteractive.com/insights/what-ai-thinks-about-your-brand-is-already-written (accessed: 05.10.2026)

²⁴ Semrush: The Most-Cited Domains in AI: A 3-Month Study (2025), https://www.semrush.com/blog/most-cited-domains-ai/ (accessed: 05.10.2026)

²⁵ Aggarwal et al.: GEO: Generative Engine Optimization, KDD 2024 (2024), https://arxiv.org/abs/2311.09735 (accessed: 05.10.2026)

²⁶ Puerto et al.: C-SEO Bench: Does Conversational SEO Work? (2025), https://arxiv.org/abs/2506.11097 (accessed: 05.10.2026)

²⁷ Search Engine Journal, Kevin Indig: The Science Of How AI Pays Attention (2026), https://www.searchenginejournal.com/the-science-of-how-ai-pays-attention/ (accessed: 05.10.2026)

²⁸ Lily Ray: Why Calling Yourself the „Best“ Could Be Helping Your Competitors Win in AI Search (2026), https://lilyraynyc.substack.com/p/why-calling-yourself-the-best-could (accessed: 05.10.2026)

²⁹ Ahrefs: Do Self-Promotional „Best“ Lists Boost ChatGPT Visibility? (2025), https://ahrefs.com/blog/best-lists-research/ (accessed: 05.10.2026)

³⁰ Lily Ray: Is Google Finally Cracking Down on Self-Promotional Listicles? (2026), https://lilyraynyc.substack.com/p/is-google-finally-cracking-down-on (accessed: 05.10.2026)

³¹ IHK Frankfurt am Main: Comparative advertising (n.d.), https://www.frankfurt-main.ihk.de/recht/uebersicht-alle-rechtsthemen/wettbewerbsrecht/unlauterer-wettbewerb/vergleichende-werbung/uebersicht-vergleichende-werbung-5273890 (accessed: 05.10.2026)

³² Filandrianos et al.: Bias Beware: The Impact of Cognitive Biases on LLM-Driven Product Recommendations (2025), https://arxiv.org/abs/2502.01349 (accessed: 05.10.2026)

³³ Ahrefs: Why ChatGPT Cites One Page Over Another, Study of 1.4M Prompts (2026), https://ahrefs.com/blog/why-chatgpt-cites-pages/ (accessed: 05.10.2026)

³⁴ Search Engine Land: Report on BrightEdge data on deeper pages cited in Google AI Overviews (2025), https://searchengineland.com/google-ai-overviews-citations-deep-pages-453414 (accessed: 05.10.2026)

³⁵ SparkToro: AIs Are Highly Inconsistent When Recommending Brands or Products (2026), https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/ (accessed: 05.10.2026)

³⁶ SISTRIX: AI Citation Drift: How Stable Are Sources in AI Search Results? (2026), https://www.sistrix.com/blog/ai-citation-drift-how-stable-are-sources-in-ai-search-results/ (accessed: 05.10.2026)

³⁷ SE Ranking: AI Traffic Research Study (2026), https://seranking.com/blog/ai-traffic-research-study/ (accessed: 05.10.2026)

³⁸ Kaiser, Schulze: Study on ChatGPT traffic performance in online stores, Marketing Science (2026), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5585812 (accessed: 05.10.2026)

³⁹ Schuh Keller: Wanderschuhe: Deine Fragen, unsere Antworten (n.d.), accessed October 19, 2026.

⁴⁰ Bergfreunde: Drückt der Schuh? Wanderschuhe richtig einlaufen (n.d.), accessed October 19, 2026.

⁴¹ Rasenwelt: Herbstdünger und Rasenpflege im Herbst (n.d.), accessed October 19, 2026.

⁴² Similarweb, Darrell Mordecai: What Happens After AI Recommends Your Brand? (October 14, 2026), updated analysis of “The Downstream Impact of AI Visibility”, including methodology and individual results; accessed October 19, 2026.

⁴³ Semrush, Margarita Loktionova: Why 62% of AI citations don’t lead to brand mentions [Study] (June 9, 2026), study with Kevin Indig and Growth Memo; accessed October 19, 2026.