Audiovisual patterns of emotional advertising and physiological engagement: a marketing science perspective on sport storytelling

Introduction Sport advertising increasingly relies on emotional storytelling, yet the Marketing Science literature lacks systematic, second-by-second evidence on which audiovisual patterns modulate viewers' physiological arousal in real commercials. Methodology This study maps the audiovisual grammar of sport advertising through a secondary analysis of 750 seconds of group-level electrodermal activity (EDA) synchronized with content coding of 22 nationally aired Spanish sport advertisements (2023-2025), viewed by 30 participants. EDA was decomposed into tonic (EDL) and phasic (EDR) components and expressed as bounded arousal indices (0-1); each second was coded by two independent raters ( κ = 0.83-0.94) across nine audiovisual categories. Micro-level analyses (one-way ANOVAs with minimum cell-size rules, repeated with cluster-robust standard errors clustered by spot) and macro-level correlations between feature density and spot-level arousal ( N = 22). Results Identified sound design and location subtype as modulators significant under both classical and cluster-robust inference, and shot type as significant only under cluster-robust inference; lighting, on-screen branding, and emotional elements were significant only under classical inference, and role, gender, and location category did not differentiate responses. Intimate human voices (murmurs) coincided with the highest tonic arousal values in the corpus ( M = 0.81; a single-spot feature reported as an illustrative case), and, descriptively, logo-and-text combinations sustained higher tonic arousal than text alone, consistent with dual coding theory. At the spot level, the density of extreme close-ups was the only nominal positive correlate of phasic arousal ( r = .433, p = .044 uncorrected; p = .257 after false-discovery-rate correction), whereas auditory-complexity densities showed no association with arousal; all macro-level associations are therefore reported as hypothesis-generating. Discussion Because seconds are nested within spots and many features recur in few advertisements, all effects are interpreted as associations within this corpus rather than causal effects, and no memory outcome was measured. Second-by-second EDA-content synchronization is positioned as a diagnostic, hypothesis-generating tool for locating arousal dynamics in real advertising and for selecting candidate features for controlled experimental testing.