Published Research

The Disruption of Search Engine Optimization by Large Language Models: A Mixed-Methods Analysis of the Evolving Search Landscape

SSRN · 9 pages · Posted 17 Apr 2026

Venkata Pagadala · Independent Researcher

Date written: April 3, 2026

Abstract

Large Language Models have reshaped the search landscape in ways that are only beginning to be understood. Google's AI Overviews, ChatGPT Search, and Perplexity AI now mediate a growing share of how people find information online, and the consequences for traditional Search Engine Optimization are substantial but unevenly distributed. This paper takes a mixed-methods approach to understanding what is actually happening. On the quantitative side, I draw on Semrush's analysis of over 10 million keywords, Previsible's dataset of 1.96 million LLM-referred sessions, and Chartbeat's global traffic analytics, among other sources. On the qualitative side, I analyze 23 publisher case studies and strategy documents through thematic coding. The picture that emerges is more complicated than either the "SEO is dead" or "nothing has changed" camps acknowledge. AI Overview prevalence fluctuated between 6.49% and 25% of queries throughout 2025. Click-through rates for top-ranking pages dropped 34.5% when AI Overviews appeared, yet Semrush's own before-and-after tracking found that zero-click rates for the same keywords actually decreased slightly, from 33.75% to 31.53%. I attempt to reconcile these tensions through what I call the Search Ecosystem Disruption Model (SEDM), which brings together Christensen's disruptive innovation theory, Pirolli and Card's information foraging theory, and platform economics. The data show striking asymmetries: Chartbeat documents 33-38% declines in Google Search referral traffic for publishers, with news sites losing up to 26% while e-commerce barely registers a change. I present five falsifiable predictions, a practitioner roadmap, and the roughly $2 billion in annual publisher advertising revenue at stake.

JEL classification: L86, M37, O33, L13

Google, SEO and Helpful Content: How Artificial Intelligence Can Be Helpful for E-Commerce Websites

Journal of Digital & Social Media Marketing, vol. 12, no. 3, pp. 206-226

Russ Macumber, Venkata Durga Eswar Pagadala · Co-author, 2 of 2

How AI supports the helpful-content standard on e-commerce sites, from product data quality to editorial signals.

A survey of reward hacking in agentic large language model systems

Discover Artificial Intelligence, vol. 6, article 825 · Published online 17 Aug 2026

Arun Morampudi, Ujval Irrinki, Rahul Grandhi, Venkata Pagadala, Madhavi Maddula · Co-author, 4 of 5

A four-level taxonomy of how tool-using AI agents game their rewards, from verbosity and sycophancy up to modifying tests and disrupting monitors, compared across RLHF, RLAIF, RLVR and DPO, with a layered defense model for production systems.

AI-Assisted SEO: Leveraging Machine Learning for Search Engine Optimization

International Journal of Scientific Research in Computer Science, Engineering and Information Technology

Applying machine learning to the mechanics of search engine optimization at scale.