Customer Lifetime Value (CLTV) is a crucial metric for evaluating the economic value that users bring to a business over their entire service cycle. Accurately predicting CLTV is essential for resource optimization, improving user retention, and maximizing return on investment (ROI). However, predicting CLTV remains challenging due to the inherent sparsity and long-tail distribution of customer spending behavior, particularly in payment scenarios where user decisions are highly dynamic and influenced by external factors. Existing methods attempt to alleviate these issues but struggle with embedding quality and distribution selection, limiting their effectiveness in capturing complex user behaviors. To address these challenges, we propose the Multi-Expert Adaptive Network (MEAN), a novel CLTV prediction framework that improves embedding representations and mitigates distribution-related errors. MEAN integrates a Multi-View Feature Express (MVFE) module to optimize multi-view representations through expert-driven feature extraction and a Distribution Adaptive Module (DAM) for soft distribution assignment, preventing error amplification from incorrect sub-distribution choice. Furthermore, we introduce an alignment mechanism to synergize MVFE and DAM via bi-directional probability alignment. Extensive offline experiments and real-world online A/B testing on the WeChat financial experimental platform demonstrate the effectiveness of MEAN.