How to alleviate the decision-making difficulty of consumers facing a series of cultural and creative products and assist consumers in decision-making and improve the shopping experience is an urgent problem in the field of unmanned retail. Eye tracking technology provides new opportunities for the human-computer interaction experience of unmanned retail. A single-factor smart recommendation method for vending machines based on total fixation duration was proposed by eye tracking technology. A simulation scenario of a cultural and creative product vending machine container was constructed for experimental verification. Through eye tracking data analysis and subjective interviews, the prediction of purchase intention and the recommendation effectiveness under different strategies were explored and analyzed. By the recommendation method, when three products were recommended, the probability of the recommendation hitting at least one consumers' purchase intention was 97.92%, and the probability of hitting the favorite product was 78.13%. The smart recommendation method can effectively predict consumers' purchase intention, narrow the decision-making scope to assist decision-making, and provide a research foundation for achieving smart shopping guidance for vending machines.