User Information Demand Prediction and Intelligent Communication Strategies Based on Customer Lifetime Value

Customer Lifetime Value (CLV) is widely used as an analytical perspective for understanding user behavior and supporting marketing decision‐making. However, accurately predicting user information demands across different lifecycle stages remains challenging due to the dynamic, sparse, and noisy nature of behavioral data. In this paper, we propose a sequential demand prediction framework that integrates multi‐head attention mechanisms with LSTM networks to model temporal dependencies in user behavior sequences. Rather than directly predicting CLV, the proposed method leverages lifecycle‐related behavioral patterns to infer user information demand tendencies over time. Experiments on a large‐scale real‐world dataset demonstrate that the proposed framework achieves stable and consistent performance in this challenging prediction setting, with an accuracy of 0.5293, a precision of 0.5162, a recall of 0.5194, an F 1 score of 0.5156, and an AUC‐ROC of 0.6414. These results indicate that incorporating temporal modeling and attention mechanisms can effectively support user demand inference and provide actionable insights for lifecycle‐oriented marketing strategies.