This study investigates beverage mixture optimization using the Interactive Genetic Algorithm (IGA), incorporating user preferences into evolutionary computation. By expanding the number of design variables from five to eight, we examined the effects of an increased search space on optimization efficiency. The research identified issues such as declining evaluation scores and preference reversals in follow-up assessments. To address these challenges, we adjusted the relative intensity of design variables and implemented an elitist selection strategy. Through a seven-generation evaluation experiment, we observed improvements in stability, higher average evaluation scores, and efficient convergence. Additionally, a re-evaluation experiment confirmed the robustness of the optimized solutions. The results highlight the significant impact of design variable intensity on search behavior, demonstrating that well-balanced intensity distributions improve optimization outcomes. The findings suggest that IGA is an effective tool for user-driven product design in sensory engineering applications, particularly in the food and beverage industry. Future work should explore adaptive evaluation frameworks and hybrid optimization techniques to further enhance solution stability and user satisfaction.