Peer-Reviewed

GEO Research

The academic and industry evidence base behind the LAVF framework. All papers are peer-reviewed or preprint research on Generative Engine Optimization and AI citation behavior. Each paper links directly to its arXiv or ACM source.

LAVF dimension:AuthorityCitation QualityContent StructureKnowledge CoverageAI Accessibility

Foundational Paper

Foundational · ACM SIGKDD 2024arXiv:2311.09735

GEO: Generative Engine Optimization

Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan, Ameet Deshpande

Princeton University

The rise of AI-powered search engines (e.g., BingChat, Google SGE) has ushered in a new paradigm for information access. This paper introduces Generative Engine Optimization (GEO) — methods to optimize sources for AI-powered search engines. The authors demonstrate that content changes following GEO principles can improve visibility in AI-generated responses by up to 40%.

Key Findings

  • Up to 40% improvement in AI visibility through structured content optimisation
  • Identifies authoritative sourcing, statistics inclusion, and fluency as primary GEO levers
  • Establishes that AI citation behaviour is measurable and systematically improvable
  • Introduces GEO as a discipline distinct from traditional SEO
Read on arXiv

Additional Research(6 papers)

Sep 2025arXiv:2509.08919Authority

Earned Media Bias in AI-Generated Responses

AI systems exhibit systematic bias toward earned-media sources when generating citations — brands with genuine editorial coverage are favoured over paid or owned channels.

arxiv.org/abs/2509.08919
Dec 2025arXiv:2512.09483Citation Quality

LLM Citation Bias: Sources and Patterns

Characterises citation bias patterns across major LLMs, identifying structural (formatting, schema) and topical (domain authority, recency) skews in what gets cited.

arxiv.org/abs/2512.09483
Mar 2026arXiv:2603.09296Citation Quality

Citation Failures and Source Inconsistency in LLMs

Documents 25–30% inconsistency in LLM citations across semantically equivalent queries — a fundamental reliability constraint for AI visibility measurement.

arxiv.org/abs/2603.09296
Mar 2026arXiv:2603.20213AI Accessibility

AgenticGEO: Optimizing Visibility in Agentic AI Systems

Extends the GEO framework to agentic AI workflows — multi-step reasoning, tool use, and autonomous agents introduce new citation dynamics beyond single-turn search.

arxiv.org/abs/2603.20213
Mar 2026arXiv:2603.29979Content Structure

GEO-SFE: Structural Features and Their Effect on AI Citations

Identifies specific structural content features — heading hierarchy, schema markup, list formatting, authoritative signals — that statistically increase LLM citation rates.

arxiv.org/abs/2603.29979
Apr 2026arXiv:2604.19113Content Structure

Feature-Level GEO: Granular Optimization for AI Visibility

Provides a feature-level breakdown of GEO signals, enabling targeted content optimisation for specific AI model citation patterns — building on GEO-SFE findings.

arxiv.org/abs/2604.19113

Industry Data

Measurement data from industry research firms complementing the academic literature.

89%of brands cited by AIyet only 14% are actively measuring itGoodFirms 2026
<1%AI recommendation consistencyacross equivalent queries from different usersSparkToro 2026
40%visibility improvementachievable through structured GEO content optimisationPrinceton / ACM SIGKDD 2024
25–30%citation inconsistencyin LLM citations across semantically equivalent queriesarXiv:2603.09296
48%of queries trigger AI Overviewup 58% year-over-yearBrightEdge 2026
40%re-appearance probabilitywhen a brand secures both mention and citation simultaneouslyAirOps 2026
25.7%more recent content cited by AIvs. traditional search results on the same topicAhrefs 2026
#1LinkedIn for professional queriestop AI-cited domain Nov 2025 – Feb 2026SE Ranking
6.4×citations/query at scalesites with 1.16M+ traffic; drops to 2.4× under 3K trafficSE Ranking
88%of Google AI Mode citationssourced from outside the regular SERP top resultsMoz 2026
0.334brand search correlationcorrelation between brand search volume and AI citation rateConvertMate (80M citation analysis)

How we apply this research

See how the LAVF framework operationalises these findings into measurable AI visibility scores.

View Methodology →

Papers are linked directly to their original arXiv or ACM sources. LLMs Network does not host or redistribute paper content. TL;DR summaries are our own interpretations for practitioner audiences.