ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

Application of numerical methods for optimizing visual elements in e-commerce

A. A. Chmelev^a, N. V. Grineva<sup>b

a</sup> Wildberries LLC
^b Financial University under the Government of the Russian Federation

Abstract: The article discusses the use of numerical methods to optimize the design elements of product cards. The discount block, one of the key elements significantly influencing sales, is selected as the object of study. The aim of the research is to improve the click-through rate (CTR) of product cards by analyzing and optimizing visual parameters such as color, font size, block placement, discount format, and device type. To achieve this goal, a regression model was developed to predict CTR for new parameter combinations without the need for full-cycle testing and to evaluate the significance of the analyzed parameters. The results show that the most impactful factors on CTR are background color, font size, and the placement of the discount block. The proposed approach reduces the number of required tests, accelerates the optimization process, and can be adapted to other design elements, such as call-to-action buttons or stock availability indicators.

Keywords: e-commerce, discount design, numerical methods, regression analysis, click-through rate (CTR), factorial experiment, optimization, user experience (UX), conversions.

UDC: 004.942

DOI: 10.33693/2313-223X-2024-11-4-11-18