Measuring Brand Visibility in Agentic Commerce Through Citation Share, Basket Capture, and Paid-Presence Transfer in Consumer Packaged Goods

Authors

  • Tarun Gupta Marketing, Reckitt, New Jersey, USA Author
  • Supriya Bansal E-commerce, Luxe Weavers, New Jersey, USA Author

DOI:

https://doi.org/10.47363/JMSCM/2026(5)194

Keywords:

Agentic Commerce, Generative Engine Optimization, Large Language Models, Brand Visibility, Consumer Packaged Goods, Paid Search, Measurement

Abstract

Generative assistants and shopping agents increasingly choose which consumer packaged goods (CPG) brands shoppers see, yet no published study has measured brand visibility in these habitual, repeat-purchase categories. This paper introduces a reproducible measurement framework built on five constructs: Citation Share (with a rank-weighted variant), the probability that a brand is named for a brand-free query; Basket Capture Rate, the share of agent-assembled basket items won by one manufacturer; Portfolio Halo, the gain in a manufacturer’s chance of winning a basket slot after winning the previous one; Source Provenance Ratio, the mix of cited domain classes; and the Paid-Presence Index, a count of visible sponsored placements that does not measure advertising spend. All data are simulated. The framework is demonstrated on 12,000 simulated responses (six US CPG categories, 60 anonymized brands, 300 queries, four anonymized engines, two waves eight weeks apart) calibrated to published studies; results show what the framework can detect, not the behavior of any real engine, brand or manufacturer. Visibility was more concentrated than published Figure for technology and services (mean
Gini 0.445 against 0.30), the four engines agreed on the top brand in 17.6 percent of items, and the leading manufacturer won 40.4 percent of basket slots against a share-proportional expectation of 25.1 percent. Paid presence was not significantly associated with Citation Share after controls, but equivalence to zero could not be established. For one in four brand-engine pairs, Citation Share shifted by at least 10 percentage points between waves. Workbooks, scripts and protocol are released.

Author Biographies

  • Tarun Gupta, Marketing, Reckitt, New Jersey, USA

    Tarun Gupta, Marketing, Reckitt, New Jersey, USA

  • Supriya Bansal, E-commerce, Luxe Weavers, New Jersey, USA

    E-commerce, Luxe Weavers, New Jersey, USA

Downloads

Published

2026-11-21