Research on Precision Customer Profiling Construction and Recommendation Algorithm Optimization in Digital Marketing Based on Big Data Analytics

Abstract

The challenges that businesses experience in the age of the digital economy are a lack of customer data, inability to properly identify user needs, and low conversion rates of marketing. The current paper suggests a big data-based design of building customer profiles and optimization of recommendation algorithms to conduct intelligent marketing. Initially, multi-dimensional customer data is segmented via K-means clustering, and an RFM (Recency-Frequency-Monetary) model is employed to measure the customer value. Second, a hybrid recommendation model, which combines collaborative filtering with deep learning, is built, utilizing convolutional neural networks (CNNs) to extract features and employing long short-term memory (LSTM) to capture temporal behavioral patterns. Lastly, an attention mechanism is presented to improve the accuracy of recommendations. Experimental results show that the K-means combined with the RFM model achieves a silhouette coefficient of 0.684, significantly higher than the 0.531 of K-means alone and the 0.598 of hierarchical clustering. This research provides theoretical support and technical pathways for enterprises to achieve precise marketing decisions, effectively solving the problems of customer identification and recommendation efficiency. RFID or other wireless retail-sensing data can be added to the customer profile when available.