The Wildberries team has introduced the "Attention Analysis" service for sellers, which uses artificial intelligence to evaluate the visual presentation of photos in product cards. The tool predicts which elements of an image will attract a buyer's attention faster and which may remain less noticeable.
The service analyzes the photo and generates several types of results. A heatmap shows the predicted areas of attention concentration, a visibility assessment reflects how much the product, text, and infographics stand out, and an eye-tracking trajectory demonstrates the presumed order of perceiving key elements.
The analysis uses an ML model that considers the peculiarities of eye movement. It works only with the image and does not access the camera or user behavior data. Processing one photo takes an average of about 15 seconds.
The results can be applied when comparing variants of the main image, changing the arrangement of the product and text elements. Sellers can further test hypotheses using A/B tests.
"Attention Analysis" is available in the product card photo editor on the seller portal with a "Jam" subscription. Currently, the service is being tested by individual sellers from Russia, Armenia, Kazakhstan, Belarus, Kyrgyzstan, Uzbekistan, and China.