Causal effect of catalog-enrichment interventions on Agentic Share-of-Search

Determine whether modifying a product-level catalog-enrichment signal shifts Agentic Share-of-Search (ASoS) by conducting the controlled intervention experiment specified for the system.

Background

The paper distinguishes association-based diagnosis from stronger predictive and intervention claims. The diagnostic agent currently identifies catalog signals that are correlated with visibility gaps, but the observed correlations may reflect retailer, brand, price-tier, or product-quality confounds and therefore do not establish that changing a signal will improve visibility.

The unresolved issue is whether concrete merchandising changes—such as modifying product titles, structured data, pricing, reviews, or other catalog attributes—produce a measurable causal change in ASoS. The paper proposes a controlled treatment-versus-control experiment with longitudinal, multi-platform measurement to answer this question.

References

We distinguish three progressively stronger claims: association (established in this paper), prediction (whether current signal levels forecast next-week visibility, testable with the weekly measurement panel), and intervention (whether modifying a signal shifts ASoS, which requires the controlled experiment specified under future work).

— Agentic Share-of-Search: A Multi-Agent AI System for Competitive Decision-Making in LLM-Mediated E-Commerce  (2609.11190 - Chowdhury, 10 Sep 2026) in Section 4, “Estimand, unit of analysis, and the association–prediction–intervention distinction,” p. 7