Authority locking vs. pluralistic sourcing in generative search: evidence compression beyond top-spot displacement

Cheng, Chiang-Yu
Electronic Commerce Research · Springer Nature · 2026-10-01 · DOI: 10.1007/s10660-026-10209-1

Abstract

Generative search is reshaping electronic commerce by replacing rank-ordered links with synthesized answers supported by selected citations. This study examines whether such citations merely mirror organic search rankings or operate as a distinct visibility-governance layer. We introduce a citation–baseline offset framework that compares generative citations with paired organic search results captured in the same query execution. Using a two-wave field audit of 240 product-related query instances, we test whether high-involvement decision contexts are associated with visible top-spot displacement or broader distributional compression. The findings show that high involvement does not systematically increase authority uplift or authority selection gain. Instead, it is associated with higher citation concentration and lower source diversity, especially in risk-avoidance tasks. The study demonstrates that generative citations can subtly narrow the evidence consumers see, with implications for search platforms, retailers, publishers, and public-interest audits of AI-mediated commerce.

Read at publisher →

Similar articles