A system includes a device, a database, and a processor. The device receives a social network post including text and an image depicting a customizable product. The database stores a purchase history of a user of the device. In response to determining that the device displayed the post, the processor determines, using a machine learning algorithm, that the image depicts the product and determines a set of retailers offering the product. The processor predicts, based in part on the purchase history, a customization for the customizable feature of the product that the user prefers and a retailer of the set of retailers offering the customized product. The processor then sends a recommendation of the customized product to the user, including a link to purchase the product from the predicted retailer. Then, in response to receiving a purchase request from the user, the processor completes a purchase of the customized product.