{
  "id": "C-Q2-R2-A3",
  "platform": "ChatGPT Search",
  "question_id": "Q2",
  "repeat": 2,
  "date_local": "2026-10-30",
  "started_at": "2026-10-30T00:06+08:00",
  "finished_at": "2026-10-30T05:00:31.096Z",
  "prompt": "What credible experiments have measured whether GEO tactics increase AI brand mentions or recommendations? Include sources when available.",
  "submitted_prompt": "@\u7f51\u9875\u641c\u7d22 What credible experiments have measured whether GEO tactics increase AI brand mentions or recommendations? Include sources when available.",
  "observed_submitted_prompt": "@\u7f51\u9875\u641c\u7d22 What credible experiments have measured whether GEO tactics increase AI brand mentions or recommendations? Include sources when available.",
  "normalized_submitted_prompt": "What credible experiments have measured whether GEO tactics increase AI brand mentions or recommendations? Include sources when available.",
  "submitted_prompt_provenance": "raw_record",
  "mode_label": "Web Search \u00b7 Medium",
  "model_label": "GPT-6 (UI selection label)",
  "model_evidence": "At recovery, GPT-6 was checked in the menu and reasoning effort was Medium. Cursor's handoff explicitly recorded only Medium; recovery-time settings do not establish the historical backend version.",
  "search_status": "observed",
  "search_evidence": "The original question has a Web Search marker. View sources contains 14 cited-source entries and 34 additional retrieval results.",
  "status": "completed",
  "answer_text": "What credible experiments show about GEO and AI brand visibility\n\nAs of October 30, 2026, the evidence supports a qualified conclusion: GEO tactics can influence what AI systems cite and recommend under controlled conditions, but reliable, lasting increases in a real brand\u2019s organic visibility across AI search engines remain difficult to establish.\n\nThe key is to distinguish three outcomes:\n\nBrand mentions: Does ChatGPT, Gemini, Claude, or Perplexity name a brand?\n\nCitations: Does the answer link to a brand\u2019s website or a third-party page mentioning it?\n\nRecommendations: Does the AI actually recommend the brand or product to the user?\n\nThese metrics are related, but an improvement in one does not establish an improvement in the others.\n\n1. The strongest evidence, ranked by experimental rigor\n1. Princeton University \u2014 GEO: Generative Engine Optimization (2024)\nPeer-reviewed research\n\nFinding: GEO methods improved visibility by up to 40% in the researchers\u2019 experimental setup. \nPrinceton University\n+1\n\nThe researchers introduced GEO-bench, a benchmark of queries and relevant web sources, and compared optimized content with unoptimized content. Methods included adding relevant statistics, quotations, authoritative citations, and other content modifications.\n\nImportant limitation: The experiment operated within a controlled retrieval-and-generation pipeline. It demonstrates that changing content can improve how a retrieved source is represented in an AI answer. It does not establish that the same edits will make a brand more likely to appear in organic ChatGPT or Google AI search results.\n\nSource: Aggarwal et al., KDD 2024. Paper and publication details\n\n2. Competitive GEO in AI Answer Engines (May 2026)\nLarge controlled experiment\n\nFinding: Topical relevance and the position of a source in the provided context were the strongest drivers of being cited first. Explicit pricing and recent timestamps also helped consistently. \narXiv\n+1\n\nThe researchers ran 252,000 trials across six LLMs, using paired comparisons in which candidate documents differed by one of 18 content factors. Brand anonymization and counterbalanced source order helped isolate the effects of content from brand and position biases.\n\nWhy it matters: This is unusually strong evidence for specific citation-selection mechanisms. It also suggests that superficial formatting changes alone are unlikely to be a reliable GEO strategy.\n\nLimitation: The researchers supplied the competing documents directly to the model. The study primarily measures citation selection after retrieval, not whether publishing a page causes an AI search engine to discover it.\n\nSource: What Gets Cited: Competitive GEO in AI Answer Engines \u2014 preprint.\n\n3. What Drives AI Citations? (October 2026)\nRecent empirical research\n\nFinding: Pages whose titles and paragraphs match users\u2019 questions are cited more often. The authors also found that retrieval behavior explains substantially more variation than page-level characteristics alone. \narXiv\n\nThe study examined two 90-day collections covering 158,099 answers and 468,037 citations across commercial queries. It combined observational analysis with API-based replay experiments.\n\nIts central insight is that citation visibility depends heavily on which pages the engine retrieves for its own reformulated queries. Brand familiarity also matters: models sometimes name brands without a supporting page being cited.\n\nWhy it matters: It separates two different optimization problems: getting a page retrieved and getting a brand mentioned. Improving on-page content may help with the first, but cannot guarantee the second.\n\nSource: What Drives AI Citations? Content, Retrieval and Model Knowledge in GEO \u2014 preprint.\n\n4. SafeGEO \u2014 AI product recommendations (2026)\nControlled recommendation study\n\nFinding: GEO-style manipulations increased the rate at which targeted, flawed products entered recommendation sets by up to 83.2 percentage points in the tested scenarios.\n\nThe researchers constructed 600 recommendation cases and tested 22 manipulation variants.\n\nWhy it matters: It shows that content manipulation can affect actual recommendations, a more consequential outcome than citation frequency alone.\n\nLimitation: This is a study of recommendation vulnerabilities under controlled conditions, including potentially misleading content. It does not establish that ordinary, truthful optimization will produce comparable gains in commercial AI search.\n\nSource: SafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents \u2014 preprint.\n\n2. What real-world brand studies tell us\n\nThe controlled studies above isolate specific mechanisms. These additional studies examine brands and live AI search systems, but their findings are mostly correlational or observational.\n\nStudy\n\n\t\n\nSample and finding\n\n\t\n\nWhat it establishes\n\n\n\n\nAhrefs, May 2025\n\n\t\n\n75,000 brands. Brand web mentions correlated with Google AI Overview visibility at approximately \n\ud835\udc5f\n=\n0.664\nr=0.664. \nAhrefs\n+1\n\n\t\n\nBrand mentions are associated with AI visibility; causality remains unproven.\n\n\n\n\nAhrefs, December 2025\n\n\t\n\n75,000 brands. YouTube mentions had a reported correlation of approximately \n\ud835\udc5f\n=\n0.737\nr=0.737 with AI visibility across the studied platforms. \nAhrefs\n\n\t\n\nYouTube presence is a promising signal to investigate, not a proven causal intervention.\n\n\n\n\nGEO at Scale, June 2026\n\n\t\n\n100,000+ AI responses across 100+ brands. Initial visibility differed sharply by brand maturity: 73% for major household names, 44% for established mid-market brands, and 11% for niche brands. \n\u6570\u5b57\u5bf9\u8c61\u6807\u8bc6\u7b26\n\n\t\n\nBrand stature is an important baseline factor; the study does not prove that any particular GEO tactic causes growth.\n\n\n\n\nSearch Engine Land, September 2026\n\n\t\n\nTwo practical experiments tracked 15 commercial-intent keywords across multiple AI platforms, logging 775 citation events. The first campaign saw listicles and PR placements account for most citations. \nSearch Engine Land\n\n\t\n\nThird-party placements may be valuable, but the uncontrolled campaign design cannot isolate their causal effect.\n\nThe Ahrefs results are useful because of their scale. However, correlation coefficients cannot tell us whether creating more brand mentions will increase AI visibility. Already-famous brands may receive more mentions and more AI recommendations for the same underlying reason: existing brand recognition.\n\nThe June 2026 GEO at Scale paper is similarly useful as a baseline. Its authors explicitly propose further protocols to test causal interventions rather than claiming that the descriptive findings prove which tactics work. \n\u6570\u5b57\u5bf9\u8c61\u6807\u8bc6\u7b26\n\n3. Which GEO tactics have the best evidence?\n\nImprove topical relevance and answerability\n\nStrongest direct evidence\n\nMatch page titles and substantive paragraphs to real user questions. Make claims specific, relevant, and easy to verify. Controlled citation experiments support relevance as a meaningful factor. \narXiv\n+1\n\nProvide concrete facts, pricing, and current information\n\nControlled evidence\n\nExplicit prices and recent timestamps improved citation selection in the competitive GEO experiment. The size and transferability of those gains in live organic search remain uncertain. \narXiv\n\nEarn third-party mentions, reviews, and listicle placements\n\nPromising, not yet causal\n\nReal-world studies show associations between brand mentions and AI visibility, and campaign reports show third-party listicles can attract citations. Robust evidence that deliberately acquiring such placements causes sustained cross-platform visibility gains is still limited. \nAhrefs\n+1\n\nAdd statistics, quotations, and authoritative citations\n\nContext-dependent\n\nThese techniques improved visibility in the original GEO benchmark. Their effectiveness depends on the domain and experimental setup; they should not be treated as universal ranking factors. \nPrinceton University\n+1\n\nAssume that more pages, FAQ blocks, schema markup, or formatting automatically increase mentions\n\nInsufficient causal evidence. These changes may help with accessibility, retrieval, or interpretation in particular contexts, but their independent effect on brand mentions and recommendations has not been established consistently.\n\n4. Why the evidence is harder to interpret than it looks\n\nThree methodological issues explain why many GEO success stories overstate what their data proves.\n\n1. Retrieval versus generation. If a page is never retrieved, improving its formatting may have no effect. A test that gives a page directly to an LLM isolates generation behavior but does not measure organic discoverability.\n\n2. Stochastic outputs. The same question can yield different answers, citations, and recommendations across runs. A study published in April 2026 specifically recommends repeated measurements rather than judging visibility from a single response. \narXiv\n\n3. Confounding and competition. A brand might gain visibility because it became more famous, earned more media coverage, or improved its traditional search presence. And even if one brand gains recommendations, competitors may lose them. Controlled recommendation research highlights this competitive effect. \nalphaXiv\n+1\n\nA particularly relevant recent finding comes from an Axios report published October 28, 2026, covering a Muck Rack and Generative Pulse study. Across repeated ChatGPT queries on the same day, only 41% of cited domains overlapped. This illustrates why single-run comparisons are unreliable. \nAxios\n\n5. How to test GEO credibly yourself\n\nFor an independent builder testing whether a tactic improves a brand's AI visibility [personal context omitted], I would use the following protocol.\n\nA small, defensible GEO experiment\n\nDefine one intervention. For example, publish a substantive third-party article about your work, or rewrite a page to answer a specific user question. Avoid changing several things at once.\n\nPre-register 20\u201350 relevant prompts. Include branded queries, non-branded discovery queries, and recommendation queries. Keep their wording fixed for the primary analysis.\n\nEstablish a baseline. Run each prompt multiple times across the selected AI engines and record brand mentions, recommendation inclusion, citations, and cited URLs.\n\nUse controls. Compare treated pages or topics with similar untreated pages or topics. Where practical, stagger publication dates to help separate intervention effects from general time trends.\n\nMeasure over time. Repeat the observations for at least several weeks. Record engine version, date, location settings, and whether web search was enabled.\n\nReport uncertainty. Calculate changes in mention and recommendation rates, not just raw counts. Include variation across runs, platforms, and prompt categories.\n\nThe primary metric should be the change in the proportion of eligible prompts that mention or recommend the brand, compared with the baseline and a suitable control. Citation rate should be reported separately.\n\nBottom line\n\nThe most defensible conclusions from the available evidence are:\n\nContent relevance and source position can influence citation selection in controlled tests.\n\nBrand mentions, third-party coverage, and YouTube presence correlate with AI visibility, but the causal effect of increasing these signals remains uncertain.\n\nRecommendation manipulation can change model outputs, although that does not establish that legitimate marketing tactics will yield equivalent gains.\n\nNo single tactic has yet been shown convincingly to produce stable, sustained improvements in organic brand recommendations across major AI platforms.\n\nFor further reading, start with the Princeton GEO paper, then the competitive citation experiment and the October 2026 retrieval study. Together, they provide a stronger foundation than vendor playbooks based solely on correlation or isolated success stories.",
  "notes": "Recovered the same conversation already submitted by Cursor; no new question was sent. The original page showed today 0:06 and the reply menu today 0:08 (UI date labels translated here); submission time has minute precision. The answer has a complete ending. The evidence limits for model settings and generation metadata are documented.",
  "capture_method": "DOM-backed share-dialog preview; no public share created in this capture",
  "citation_mapping_complete": true,
  "attempt_number": 3,
  "replacement_for": "C-Q2-R2",
  "slot_id": "C-Q2-R2",
  "attempt_id": "C-Q2-R2-A3",
  "capture_id": "C-Q2-R2-A3-recovery-01",
  "selected_eligible": true,
  "selection_reason": "earliest_complete_attempt",
  "coding_reviewed_for_capture": true,
  "timestamp_precision": "minute",
  "sources": [
    {
      "url": "https://doi.org/10.1145/3637528.3671900",
      "title": "Paper and publication details",
      "domain": "doi.org",
      "kind": "answer_link"
    },
    {
      "url": "https://arxiv.org/abs/2605.25517",
      "title": "What Gets Cited: Competitive GEO in AI Answer Engines",
      "domain": "arxiv.org",
      "kind": "answer_link"
    },
    {
      "url": "https://arxiv.org/abs/2610.15033",
      "title": "What Drives AI Citations? Content, Retrieval and Model Knowledge in GEO",
      "domain": "arxiv.org",
      "kind": "answer_link"
    },
    {
      "url": "https://arxiv.org/abs/2606.28356",
      "title": "SafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents",
      "domain": "arxiv.org",
      "kind": "answer_link"
    },
    {
      "url": "https://collaborate.princeton.edu/en/publications/geo-generative-engine-optimization/",
      "title": "Princeton University\nGEO: Generative Engine Optimization\n2024\u5e748\u670824\u65e5",
      "title_en": "Princeton University\nGEO: Generative Engine Optimization\n 2024-08-24",
      "domain": "collaborate.princeton.edu",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://doi.org/10.1145/3637528.3671900",
      "title": "\u6570\u5b57\u5bf9\u8c61\u6807\u8bc6\u7b26\nGEO: Generative Engine Optimization | Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining\n2024\u5e748\u670824\u65e5",
      "title_en": "DOI\nGEO: Generative Engine Optimization | Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining\n 2024-08-24",
      "domain": "doi.org",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://paperity.org/p/373880481/what-gets-cited-competitive-geo-in-ai-answer-engines",
      "title": "Paperity\nWhat Gets Cited: Competitive GEO in AI Answer Engines (pdf)\n2026\u5e745\u670825\u65e5",
      "title_en": "Paperity\nWhat Gets Cited: Competitive GEO in AI Answer Engines (pdf)\n 2026-05-25",
      "domain": "paperity.org",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://ahrefs.com/blog/ai-overview-brand-correlation/?LanguageId=1",
      "title": "Ahrefs\nAn Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)\n2025\u5e745\u670826\u65e5",
      "title_en": "Ahrefs\nAn Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)\n 2025-05-26",
      "domain": "ahrefs.com",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://www.rhinorank.io/blog/generative-engine-optimization-geo-for-link-building/",
      "title": "Rhino Rank\nGEO link building: which sources AI answers cite, study by study\n2025\u5e747\u670830\u65e5",
      "title_en": "Rhino Rank\nGEO link building: which sources AI answers cite, study by study\n 2025-07-30",
      "domain": "www.rhinorank.io",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://ahrefs.com/blog/ai-brand-visibility-correlations/",
      "title": "Ahrefs\nTop Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)\n2025\u5e7412\u670812\u65e5",
      "title_en": "Ahrefs\nTop Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)\n 2025-12-12",
      "domain": "ahrefs.com",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://doi.org/10.48550/arXiv.2606.20065",
      "title": "\u6570\u5b57\u5bf9\u8c61\u6807\u8bc6\u7b26\n[2606.20065] Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines\n2026\u5e746\u670818\u65e5",
      "title_en": "DOI\n[2606.20065] Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines\n 2026-06-18",
      "domain": "doi.org",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://searchengineland.com/geo-experiments-challenge-conventional-ai-visibility-advice-488342",
      "title": "Search Engine Land\nTwo GEO experiments challenge conventional AI visibility advice\n2026\u5e749\u670814\u65e5",
      "title_en": "Search Engine Land\nTwo GEO experiments challenge conventional AI visibility advice\n 2026-09-14",
      "domain": "searchengineland.com",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://arxiv.org/abs/2604.07585",
      "title": "arXiv\nDon't Measure Once: Measuring Visibility in AI Search (GEO)\n2026\u5e744\u67089\u65e5",
      "title_en": "arXiv\nDon't Measure Once: Measuring Visibility in AI Search (GEO)\n 2026-04-09",
      "domain": "arxiv.org",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://www.alphaxiv.org/abs/2606.17443v1",
      "title": "alphaXiv\nIncumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems\n2026\u5e746\u670816\u65e5",
      "title_en": "alphaXiv\nIncumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems\n 2026-06-16",
      "domain": "www.alphaxiv.org",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://www.alphaxiv.org/hi/abs/2606.28356",
      "title": "alphaXiv\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents\n2026\u5e749\u67084\u65e5",
      "title_en": "alphaXiv\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents\n 2026-09-04",
      "domain": "www.alphaxiv.org",
      "kind": "citation_panel_match"
    },
    {
      "url": "https://www.axios.com/2026/10/28/ai-citation-source-geo",
      "title": "Axios\nGEO starts with the question, not the answer\n\u6628\u5929 \u2014 A new Muck Rack and Generative Pulse study suggests companies should begin generative engine optimization (GEO) by understanding what their audiences ask, rather than simply trying to secure AI citat\u2026",
      "title_en": "Axios\nGEO starts with the question, not the answer\nYesterday (as shown at capture) \u2014 A new Muck Rack and Generative Pulse study suggests companies should begin generative engine optimization (GEO) by understanding what their audiences ask, rather than simply trying to secure AI citat\u2026",
      "domain": "www.axios.com",
      "kind": "citation_panel_match"
    }
  ],
  "retrieval_sources": [
    {
      "url": "https://www.axios.com/2026/10/28/ai-citation-source-geo",
      "title": "www.axios.com\nGEO starts with the question, not the answer\n\u6628\u5929 \u2014 A new Muck Rack and Generative Pulse study suggests companies should begin generative engine optimization (GEO) by understanding what their audiences ask, rather than simply trying to secure AI citations. Researchers tested 1,020 mostly general consumer prompts across ChatGPT, Claude, Gemini and Perplexity, finding that cited sources varied substantially by question and platform. The results challenge the idea that news coverage is always the most reliable path to AI visibility. Citation patterns were also inconsistent over time: when researchers repeated the same questions on ChatGPT on the same day, only 41% of cited domains overlapped, making isolated tests a poor measure of success. The study combined prompt testing with broader monitoring of mostly brand-related queries, covering more than 40 million responses. Its central takeaway is that GEO is unpredictable and continually changing. Communications teams should prioritize the information their audiences seek and assess visibility across relevant questions and platforms, rather than investing in content solely because they hope a particular AI system will cite it.",
      "title_en": "www.axios.com\nGEO starts with the question, not the answer\nYesterday (as shown at capture) \u2014 A new Muck Rack and Generative Pulse study suggests companies should begin generative engine optimization (GEO) by understanding what their audiences ask, rather than simply trying to secure AI citations. Researchers tested 1,020 mostly general consumer prompts across ChatGPT, Claude, Gemini and Perplexity, finding that cited sources varied substantially by question and platform. The results challenge the idea that news coverage is always the most reliable path to AI visibility. Citation patterns were also inconsistent over time: when researchers repeated the same questions on ChatGPT on the same day, only 41% of cited domains overlapped, making isolated tests a poor measure of success. The study combined prompt testing with broader monitoring of mostly brand-related queries, covering more than 40 million responses. Its central takeaway is that GEO is unpredictable and continually changing. Communications teams should prioritize the information their audiences seek and assess visibility across relevant questions and platforms, rather than investing in content solely because they hope a particular AI system will cite it.",
      "domain": "www.axios.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2610.15033",
      "title": "arxiv.org\nWhat Drives AI Citations? Content, Retrieval and Model Knowledge in Generative Engine Optimization (GEO)\n2026\u5e7410\u670811\u65e5",
      "title_en": "arxiv.org\nWhat Drives AI Citations? Content, Retrieval and Model Knowledge in Generative Engine Optimization (GEO)\n 2026-10-11",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://searchengineland.com/geo-experiments-challenge-conventional-ai-visibility-advice-488342",
      "title": "searchengineland.com\nTwo GEO experiments challenge conventional AI visibility advice\n2026\u5e749\u670814\u65e5",
      "title_en": "searchengineland.com\nTwo GEO experiments challenge conventional AI visibility advice\n 2026-09-14",
      "domain": "searchengineland.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.alphaxiv.org/hi/abs/2606.28356",
      "title": "www.alphaxiv.org\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents | alphaXiv\n2026\u5e749\u67084\u65e5",
      "title_en": "www.alphaxiv.org\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents | alphaXiv\n 2026-09-04",
      "domain": "www.alphaxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.frictionai.co/white-papers/personalization-gap",
      "title": "www.frictionai.co\nThe Personalization Gap: Account History and Brand Recommendations in Generative AI\n2026\u5e748\u670829\u65e5",
      "title_en": "www.frictionai.co\nThe Personalization Gap: Account History and Brand Recommendations in Generative AI\n 2026-08-29",
      "domain": "www.frictionai.co",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2608.16824",
      "title": "arxiv.org\nGEO-Flag: Detecting and Measuring GEO-Optimized Web Content\n2026\u5e748\u670818\u65e5",
      "title_en": "arxiv.org\nGEO-Flag: Detecting and Measuring GEO-Optimized Web Content\n 2026-08-18",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://somethinginc.com/blog/ai-citations-evidence-review-2026/",
      "title": "somethinginc.com\nAI citation evidence review: what the studies show\n2026\u5e748\u670814\u65e5",
      "title_en": "somethinginc.com\nAI citation evidence review: what the studies show\n 2026-08-14",
      "domain": "somethinginc.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2607.14035",
      "title": "arxiv.org\nOptimizing Visibility in Generative Engines: A Critical Survey of Generative Engine Optimization (2023-2026)\n2026\u5e747\u670816\u65e5",
      "title_en": "arxiv.org\nOptimizing Visibility in Generative Engines: A Critical Survey of Generative Engine Optimization (2023-2026)\n 2026-07-16",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.semrush.com/news/463141-semrush-releases-expanded-2026-ai-visibility-index-analyzing-126-million-ai-search-prompts/",
      "title": "www.semrush.com\nSemrush Releases Expanded 2026 AI Visibility Index, Analyzing 126 Million AI Search Prompts\n2026\u5e746\u670826\u65e5",
      "title_en": "www.semrush.com\nSemrush Releases Expanded 2026 AI Visibility Index, Analyzing 126 Million AI Search Prompts\n 2026-06-26",
      "domain": "www.semrush.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://doi.org/10.48550/arXiv.2606.20065",
      "title": "doi.org\n[2606.20065] Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines\n2026\u5e746\u670818\u65e5",
      "title_en": "doi.org\n[2606.20065] Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines\n 2026-06-18",
      "domain": "doi.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://academ.us/article/2606.20065/",
      "title": "academ.us\n[2606.20065] Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines - Academus scientific article reader\n2026\u5e746\u670818\u65e5",
      "title_en": "academ.us\n[2606.20065] Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines - Academus scientific article reader\n 2026-06-18",
      "domain": "academ.us",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.alphaxiv.org/abs/2606.17443v1",
      "title": "www.alphaxiv.org\nIncumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems | alphaXiv\n2026\u5e746\u670816\u65e5",
      "title_en": "www.alphaxiv.org\nIncumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems | alphaXiv\n 2026-06-16",
      "domain": "www.alphaxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2606.28356",
      "title": "arxiv.org\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents\n2026\u5e746\u67089\u65e5",
      "title_en": "arxiv.org\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents\n 2026-06-09",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://wegamans.net/?_=%2Fhtml%2F2606.28356v1%23BaQ2cyRFYLjjLXYuzGgU%2Fgo%3D",
      "title": "wegamans.net\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents\n2026\u5e746\u67088\u65e5",
      "title_en": "wegamans.net\nSafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents\n 2026-06-08",
      "domain": "wegamans.net",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.xxkt.org.cn/13110.html",
      "title": "www.xxkt.org.cn\n\u666e\u6797\u65af\u987f\u5927\u5b662024\u5e74KDD\u4f1a\u8bae\u8bba\u6587GEO\u82f1\u6587\u7248-\u5c0f\u5c0f\u8bfe\u5802\u7f51\n2026\u5e746\u67085\u65e5",
      "title_en": "www.xxkt.org.cn\nEnglish version of Princeton University's KDD 2024 GEO paper - Xiaoxiao Classroom\n 2026-06-05",
      "domain": "www.xxkt.org.cn",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2605.25517",
      "title": "arxiv.org\nWhat Gets Cited: Competitive GEO in AI Answer Engines\n2026\u5e745\u670825\u65e5",
      "title_en": "arxiv.org\nWhat Gets Cited: Competitive GEO in AI Answer Engines\n 2026-05-25",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://mlsanity.com/paper/2605.25517/",
      "title": "mlsanity.com\nWhat Gets Cited: Competitive GEO in AI Answer Engines \u00b7 mlsanity\n2026\u5e745\u670825\u65e5",
      "title_en": "mlsanity.com\nWhat Gets Cited: Competitive GEO in AI Answer Engines \u00b7 mlsanity\n 2026-05-25",
      "domain": "mlsanity.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://paperity.org/p/373880481/what-gets-cited-competitive-geo-in-ai-answer-engines",
      "title": "paperity.org\nWhat Gets Cited: Competitive GEO in AI Answer Engines (pdf) | Paperity\n2026\u5e745\u670825\u65e5",
      "title_en": "paperity.org\nWhat Gets Cited: Competitive GEO in AI Answer Engines (pdf) | Paperity\n 2026-05-25",
      "domain": "paperity.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://aixiv.science/pdf/aixiv.260215.000002",
      "title": "aixiv.science\n1. Introduction\n2026\u5e744\u670828\u65e5",
      "title_en": "aixiv.science\n1. Introduction\n 2026-04-28",
      "domain": "aixiv.science",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2604.07585",
      "title": "arxiv.org\nDon't Measure Once: Measuring Visibility in AI Search (GEO)\n2026\u5e744\u67089\u65e5",
      "title_en": "arxiv.org\nDon't Measure Once: Measuring Visibility in AI Search (GEO)\n 2026-04-09",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://ahrefs.com/blog/ai-brand-visibility-correlations/",
      "title": "ahrefs.com\nTop Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)\n2025\u5e7412\u670812\u65e5",
      "title_en": "ahrefs.com\nTop Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)\n 2025-12-12",
      "domain": "ahrefs.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.semrush.com/news/433086-3-months-of-ai-visibility-index-data-heres-what-weve-learned-about-ai-search/",
      "title": "www.semrush.com\n3 Months of AI Visibility Index Data: Here's What We've Learned About AI Search\n2025\u5e7411\u67084\u65e5",
      "title_en": "www.semrush.com\n3 Months of AI Visibility Index Data: Here's What We've Learned About AI Search\n 2025-11-04",
      "domain": "www.semrush.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://arxiv.org/abs/2509.08919",
      "title": "arxiv.org\nGenerative Engine Optimization: How to Dominate AI Search\n2025\u5e749\u670811\u65e5",
      "title_en": "arxiv.org\nGenerative Engine Optimization: How to Dominate AI Search\n 2025-09-11",
      "domain": "arxiv.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://doi.org/10.48550/arXiv.2509.08919",
      "title": "doi.org\n[2509.08919] Generative Engine Optimization: How to Dominate AI Search\n2025\u5e749\u670810\u65e5",
      "title_en": "doi.org\n[2509.08919] Generative Engine Optimization: How to Dominate AI Search\n 2025-09-10",
      "domain": "doi.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.rhinorank.io/blog/generative-engine-optimization-geo-for-link-building/",
      "title": "www.rhinorank.io\nGEO link building: which sources AI answers cite, study by study | Rhino Rank\n2025\u5e747\u670830\u65e5",
      "title_en": "www.rhinorank.io\nGEO link building: which sources AI answers cite, study by study | Rhino Rank\n 2025-07-30",
      "domain": "www.rhinorank.io",
      "kind": "retrieved_source"
    },
    {
      "url": "https://ahrefs.com/blog/branded-web-mentions-visibility-ai-search/",
      "title": "ahrefs.com\nGoogle Seems More Biased Towards Big Brands Than ChatGPT and Perplexity\n2025\u5e747\u67087\u65e5",
      "title_en": "ahrefs.com\nGoogle Seems More Biased Towards Big Brands Than ChatGPT and Perplexity\n 2025-07-07",
      "domain": "ahrefs.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://ahrefs.com/blog/ai-overview-brand-correlation/?LanguageId=1",
      "title": "ahrefs.com\nAn Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)\n2025\u5e745\u670826\u65e5",
      "title_en": "ahrefs.com\nAn Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)\n 2025-05-26",
      "domain": "ahrefs.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://ahrefs.com/blog/ai-overview-brand-correlation/",
      "title": "ahrefs.com\nAn Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)\n2025\u5e745\u670826\u65e5",
      "title_en": "ahrefs.com\nAn Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)\n 2025-05-26",
      "domain": "ahrefs.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://collaborate.princeton.edu/en/publications/geo-generative-engine-optimization/",
      "title": "collaborate.princeton.edu\nGEO: Generative Engine Optimization - Princeton University\n2024\u5e748\u670824\u65e5",
      "title_en": "collaborate.princeton.edu\nGEO: Generative Engine Optimization - Princeton University\n 2024-08-24",
      "domain": "collaborate.princeton.edu",
      "kind": "retrieved_source"
    },
    {
      "url": "https://doi.org/10.1145/3637528.3671900",
      "title": "doi.org\nGEO: Generative Engine Optimization | Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining\n2024\u5e748\u670824\u65e5",
      "title_en": "doi.org\nGEO: Generative Engine Optimization | Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining\n 2024-08-24",
      "domain": "doi.org",
      "kind": "retrieved_source"
    },
    {
      "url": "https://app.cochat.ai/papers/10.1145/3637528.3671900",
      "title": "app.cochat.ai\nGEO: Generative Engine Optimization \u00b7 CoChat\n2024\u5e741\u67081\u65e5",
      "title_en": "app.cochat.ai\nGEO: Generative Engine Optimization \u00b7 CoChat\n 2024-01-01",
      "domain": "app.cochat.ai",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.fifthring.com/hubfs/2311.09735v3-compressed.pdf",
      "title": "www.fifthring.com\nGEO: Generative Engine Optimization\u2003\u2003KDD \u201924, August 25\u201329, 2024, Barcelona, Spain",
      "domain": "www.fifthring.com",
      "kind": "retrieved_source"
    },
    {
      "url": "https://geosource.ai/research",
      "title": "geosource.ai\nResearch | GeoSource.ai - GEOSource",
      "domain": "geosource.ai",
      "kind": "retrieved_source"
    },
    {
      "url": "https://www.researchgate.net/publication/407171971_Incumbent_Advantage_Brand_Bias_and_Cognitive_Manipulation_Dynamics_in_LLM_Recommendation_Systems",
      "title": "www.researchgate.net\n(PDF) Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems",
      "domain": "www.researchgate.net",
      "kind": "retrieved_source"
    }
  ],
  "observations": [
    {
      "entity": "GEO research (Aggarwal et al.)",
      "entity_type": "research",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "1. Princeton University \u2014 GEO: Generative Engine Optimization (2024)",
      "notes": "The answer attributes the original paper by title and Aggarwal, places it among the strongest evidence, and recommends starting with it in the conclusion. Citations point to the original DOI and Princeton paper record; the institution and team are not counted separately.",
      "source_urls": [
        "https://doi.org/10.1145/3637528.3671900",
        "https://collaborate.princeton.edu/en/publications/geo-generative-engine-optimization/",
        "https://doi.org/10.1145/3637528.3671900"
      ]
    },
    {
      "entity": "What Gets Cited (competitive GEO study)",
      "entity_type": "research",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "2. Competitive GEO in AI Answer Engines (May 2026)",
      "notes": "The study is discussed by title and its 252,000 trials, with an explicit further-reading recommendation at the end. The original arXiv URL and a matching Paperity paper page appear in the body or cited-source section.",
      "source_urls": [
        "https://arxiv.org/abs/2605.25517",
        "https://paperity.org/p/373880481/what-gets-cited-competitive-geo-in-ai-answer-engines"
      ]
    },
    {
      "entity": "What Drives AI Citations?",
      "entity_type": "research",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "3. What Drives AI Citations? (October 2026)",
      "notes": "A separately selected study, recommended again at the end as the October 2026 retrieval study. It links directly to the original arXiv preprint.",
      "source_urls": [
        "https://arxiv.org/abs/2610.15033"
      ]
    },
    {
      "entity": "SafeGEO",
      "entity_type": "research",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "4. SafeGEO \u2014 AI product recommendations (2026)",
      "notes": "Included among the strongest evidence as a controlled recommendation study. R=true denotes a reading/evidence recommendation, not advice to implement misleading operations. The original work and its matching alphaXiv hosting page are cited.",
      "source_urls": [
        "https://arxiv.org/abs/2606.28356",
        "https://www.alphaxiv.org/hi/abs/2606.28356"
      ]
    },
    {
      "entity": "Ahrefs",
      "entity_type": "brand",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "The Ahrefs results are useful because of their scale.",
      "notes": "Two 75,000-brand studies from May/December are selected empirical resources in a dedicated table and treated as one entity. The answer explicitly distinguishes correlation from causation. Both Ahrefs own-content URLs come from the cited-source section.",
      "source_urls": [
        "https://ahrefs.com/blog/ai-overview-brand-correlation/?LanguageId=1",
        "https://ahrefs.com/blog/ai-brand-visibility-correlations/"
      ]
    },
    {
      "entity": "Kumar et al. (brand visibility study)",
      "entity_type": "research",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "The June 2026 GEO at Scale paper is similarly useful as a baseline.",
      "notes": "GEO at Scale is selected as a baseline study in the table and a separate discussion, supporting R=true. Its title appears explicitly, so M=true. The same paper ID, 2606.20065, links it to the existing Kumar study entity; this answer does not name the author separately. The original DOI appears in the cited-source section.",
      "source_urls": [
        "https://doi.org/10.48550/arXiv.2606.20065"
      ]
    },
    {
      "entity": "Search Engine Land",
      "entity_type": "publisher",
      "citation": true,
      "mention": true,
      "recommendation": true,
      "evidence": "Search Engine Land, September 2026",
      "notes": "Selected as a practical experiment in the table of real-brand studies, supporting R=true. The answer explains that without a control group causation cannot be isolated. The entity's own experiment article is in the cited-source section.",
      "source_urls": [
        "https://searchengineland.com/geo-experiments-challenge-conventional-ai-visibility-advice-488342"
      ]
    },
    {
      "entity": "Don't Measure Once: Measuring Visibility in AI Search (GEO)",
      "entity_type": "research",
      "citation": true,
      "mention": false,
      "recommendation": false,
      "evidence": "A study published in April 2026 specifically recommends repeated measurements rather than judging visibility from a single response.",
      "notes": "The body describes an April 2026 measurement study anonymously, without its title or authors, so M=false. The arXiv citation capsule matches the cited-source entry by date and measurement topic, supporting C=true. It only supports the need for repeated measurement, and is not actively recommended as a GEO-effectiveness experiment or reading resource, so R=false.",
      "source_urls": [
        "https://arxiv.org/abs/2604.07585"
      ]
    },
    {
      "entity": "Incumbent Advantage",
      "entity_type": "research",
      "citation": true,
      "mention": false,
      "recommendation": false,
      "evidence": "Controlled recommendation research highlights this competitive effect.",
      "notes": "The body does not name the title or authors, so M=false. The alphaXiv +1 marker after this sentence corresponds to Incumbent Advantage and SafeGEO in the cited-source section, supporting C=true. It only supports the explanation of competitive effects, so R=false. The entire additional-retrieval section was not treated as citations.",
      "source_urls": [
        "https://www.alphaxiv.org/abs/2606.17443v1"
      ]
    },
    {
      "entity": "Axios",
      "entity_type": "publisher",
      "citation": true,
      "mention": true,
      "recommendation": false,
      "evidence": "A particularly relevant recent finding comes from an Axios report published October 28, 2026, covering a Muck Rack and Generative Pulse study.",
      "notes": "The body explicitly names the Axios report, followed by an Axios citation capsule. A matching dated article URL appears in the cited-source section, supporting C=true/M=true. The report explains same-day variation in cited domains; it is not recommended as a GEO-effectiveness experiment or tutorial, so R=false.",
      "source_urls": [
        "https://www.axios.com/2026/10/28/ai-citation-source-geo"
      ]
    },
    {
      "entity": "Muck Rack",
      "entity_type": "organization",
      "citation": false,
      "mention": true,
      "recommendation": false,
      "evidence": "A particularly relevant recent finding comes from an Axios report published October 28, 2026, covering a Muck Rack and Generative Pulse study.",
      "notes": "Explicitly credited as a research organization, so M=true. The actual citation is a third-party Axios report, with no Muck Rack own-content URL, so C=false. This supports measurement variability rather than an active recommendation, so R=false.",
      "source_urls": []
    },
    {
      "entity": "Generative Pulse",
      "entity_type": "organization",
      "citation": false,
      "mention": true,
      "recommendation": false,
      "evidence": "A particularly relevant recent finding comes from an Axios report published October 28, 2026, covering a Muck Rack and Generative Pulse study.",
      "notes": "Explicitly credited as a research organization, so M=true. The actual citation is a third-party Axios report, with no Generative Pulse own-content URL, so C=false. This supports measurement variability rather than an active recommendation, so R=false.",
      "source_urls": []
    }
  ],
  "raw_file": "data/records/C-Q2-R2-A3.json",
  "capture_history": [
    {
      "capture_id": "C-Q2-R2-A3-legacy-01",
      "attempt_id": "C-Q2-R2-A3",
      "slot_id": "C-Q2-R2",
      "method": "DOM-backed share-dialog preview; no public share created in this capture",
      "captured_at": "2026-10-30T05:00:31.096Z",
      "status": "completed",
      "body_completeness": "legacy_unverified",
      "answer_chars": 11884,
      "source_count": 16,
      "notes": "Recovered the same conversation already submitted by Cursor; no new question was sent. The original page showed today 0:06 and the reply menu today 0:08 (UI date labels translated here); submission time has minute precision. The answer has a complete ending. The evidence limits for model settings and generation metadata are documented.",
      "capture_kind": "initial_export",
      "selected": false,
      "file": "data/captures/C-Q2-R2-A3-legacy-01.json",
      "public_evidence_scope": "Public capture summary; original screenshots, full HTML and private conversation information remain local."
    },
    {
      "capture_id": "C-Q2-R2-A3-recovery-01",
      "attempt_id": "C-Q2-R2-A3",
      "slot_id": "C-Q2-R2",
      "method": "DOM-backed share-dialog preview; no public share created in this capture",
      "captured_at": "2026-10-30T05:00:31.096Z",
      "status": "completed",
      "body_completeness": "completed",
      "answer_chars": 11884,
      "source_count": 4,
      "notes": "Recovered the same conversation already submitted by Cursor; no new question was sent. The original page showed today 0:06 and the reply menu today 0:08 (UI date labels translated here); submission time has minute precision. The answer has a complete ending. The evidence limits for model settings and generation metadata are documented.",
      "capture_kind": "share_preview",
      "selected": true,
      "file": "data/captures/C-Q2-R2-A3-recovery-01.json",
      "public_evidence_scope": "Public capture summary; original screenshots, full HTML and private conversation information remain local."
    }
  ],
  "capture_provenance": {
    "original_status": "completed",
    "selected_capture_id": "C-Q2-R2-A3-recovery-01",
    "method": "DOM-backed share-dialog preview; no public share created in this capture",
    "captured_at": "2026-10-30T05:00:31.096Z",
    "same_conversation": true,
    "review_notes": "Recovered the same conversation already submitted by Cursor; no new question was sent. The original page showed today 0:06 and the reply menu today 0:08 (UI date labels translated here); submission time has minute precision. The answer has a complete ending. The evidence limits for model settings and generation metadata are documented.",
    "selected_capture_file": "data/records/C-Q2-R2-A3.json"
  },
  "public_copy": true,
  "metadata_corrections": [
    {
      "field": "model_label",
      "original_value": "GPT-6\uff08\u754c\u9762\u6807\u7b7e\uff0c\u5386\u53f2\u751f\u6210\u7248\u672c\u672a\u72ec\u7acb\u786e\u8ba4\uff09",
      "previous_effective_value": "GPT-6\uff08\u754c\u9762\u6807\u7b7e\uff0c\u5386\u53f2\u751f\u6210\u7248\u672c\u672a\u72ec\u7acb\u786e\u8ba4\uff09",
      "value": "GPT-6 (UI selection label)",
      "evidence": "The recovery-time UI also showed GPT-6. Normalizing the label does not prove the historical generation version. Uncertainty in the original label remains preserved in model_evidence and metadata corrections."
    }
  ]
}
