A Multi-Factor Brand-Recognition Audit for AI Answer Engines: An Open Instrument and Protocol for Measuring Generative Visibility
Year: 2023
Type: preprint
Abstract: As AI search engines (ChatGPT, Claude, Perplexity, Google AI Overview) supplant traditional ranked-results SERPs with conversational answers that cite individual sources, the optimization target for online retailers is no longer "rank in position one" but "be cited in the model's response." We present an empirical case study of applying Generative Engine Optimization (GEO) to a 2,766-SKU direct-to-consumer women's fashion catalog (Livostyle.com, operated by Arcada LLC) over a 6-month period. We document a 14-track strategy combining: (1) structured data publication; (2) machine-readable open-data artifacts; (3) an MCP server published to npm and the Glama registry; and (4) public shareable Claude conversations demonstrating the resulting recommendation quality. We report initial implementation metrics (4.76 mean product rating, 15,937 reviews, 99% review coverage), discuss costs (~$95 in marginal infrastructure spend), and outline a measurement framework for AI-citation-rate as the new primary KPI. The artifacts are MIT-licensed and reproducible. We argue that for sub-$10M-revenue DTC retailers, GEO offers a 12-24 month window of structural advantage before large marketplaces close the gap through scale.
Source: arXiv (Cornell University)
Authors: Aggarwal, Pranjal, Murahari, Vishvak, Rajpurohit, Tanmay, Kalyan, Ashwin, Narasimhan, Karthik, Deshpande, Ameet
Institutions: Princeton University, Indian Institute of Technology Delhi
Language: English
Subfield: Sociology and Political Science
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
Funder: National Science Foundation
OpenAlex ID: W4388788661