Disclosure Specificity, Demonstration Depth, and Viewer Momentum in Social Commerce Streams: Evidence from a Staggered Platform Panel

Abstract

Social commerce compresses persuasion, information exchange, and purchase execution into the same short time window. Buyers observe a host’s spoken claims, visible product demonstrations, pinned sponsorship language, real-time audience reactions, and early transaction flow before deciding whether to purchase. These cues are all market signals, yet they differ in origin, timing, and credibility. Some are generated by the seller, some by the platform, and some by other buyers. The present study examines how these distinct signals shape commercial outcomes in live-stream shopping. The empirical setting is a calibrated panel of 31,284 creator--product stream sessions linked to 2,964,511 order records, refund requests, and post-stream repeat-purchase behavior. Two platform changes structure the analysis: a staggered rollout of mandatory pinned disclosure templates and a separate change in the placement of real-time purchase counters. The design makes it possible to estimate how disclosure detail, demonstration depth, and viewer-generated momentum affect both immediate sales and post-purchase outcomes. Stream-level fixed-effects models, stacked event studies, survival models for time to early sales milestones, and order-level refund regressions are used. Moderate increases in disclosure specificity improve checkout conversion and reduce refunds, but the effect is nonlinear and becomes weaker for low-fit creator--category pairings. Demonstration depth has a strong positive association with sales per viewer and subsequent repeat purchase. Viewer momentum raises orders quickly, but its contribution is least durable when product demonstration is thin. After the disclosure-template rollout, gross sales dip slightly for two weeks, then recover, while refunds and service contacts decline. The findings indicate that signal source and temporal sequence matter: seller-controlled cues help most when they are concrete, platform-mandated disclosures improve calibration, and crowd-generated momentum is most effective when anchored by substantive product evidence