Customer loyalty in quick commerce (q-commerce) presents unique challenges due to ultrafast delivery expectations, frequent low-value transactions, and intense competition. Traditional loyalty prediction models often fail to address these dynamics. This study evaluates machine learning approaches to predict customer loyalty in q-commerce, comparing five algorithms using real-world transactional data. Among the tested models, Random Forest demonstrates superior performance, effectively capturing complex behavioural patterns while minimizing misclassifications. Key drivers of loyalty are identified, including delivery reliability, order frequency, and promotional engagement. The findings enable businesses to strategically segment customers, optimize retention efforts, and prioritize operational improvements. This research contributes a data-driven framework for loyalty prediction specifically adapted to q-commerce environments, offering both academic and industry value in an increasingly competitive digital marketplace.