Visual scanning pattern of copycat brand packaging: a decision tree algorithm-based classification for machine learning

Purpose This study aims to develop a machine learning model to distinguish market leaders from copycat packaging by analyzing eye-tracking data, focusing on visual scanning patterns. It explores how these patterns influence consumer decision-making, integrating signaling theory to understand the impact of visual similarities on brand perception. Design/methodology/approach Eye-tracking data from 273 participants analyzing 2,043 packaging images was used to train and evaluate decision 3 models. Key visual scanning variables, such as time to first fixation, fixation count, fixations before and visit count, were analyzed to identify patterns distinguishing leader from copycat packaging effectively. Additionally, a logistic regression benchmark was performed to validate the robustness of the classification outcomes, confirming that the predictive results were not model-specific. Findings The simplified decision tree model demonstrated superior accuracy and interpretability, effectively distinguishing leader from copycat packaging. Key variables, especially visit count and time to first fixation, were critical in reducing consumer confusion, highlighting the importance of distinct visual signals in packaging design. Originality/value This study uniquely integrates eye-tracking data and signaling theory within a machine learning framework to analyze copycat packaging. Its approach offers practical insights for marketers and regulators on enhancing brand differentiation and minimizing the impact of deceptive packaging.