This research presents an advanced framework for the collection and preprocessing of social media data from various platforms, designed to enhance business decision-making. The system leverages multiple application programming interfaces (APIs) and web scraping techniques to systematically gather data from platforms such as Twitter, Facebook, Instagram, Reddit, and YouTube. The collected data undergoes a rigorous preprocessing pipeline that includes cleaning, normalization, tokenization, and feature extraction using advanced Natural Language Processing (NLP) and machine learning models. This preprocessing stage ensures data consistency, reduces noise, and enriches the data with sentiment analysis, named entity recognition, and topic modeming. The framework employs scalable technologies like Apache Kafka for real-time data streaming and Apache Spark for large-scale data processing. The processed data is stored in a NoSQL database, ensuring efficient retrieval and analysis. The framework's effectiveness is validated through several case studies demonstrating significant improvements in data quality and analysis efficiency, leading to actionable business insights. Results highlight the framework's capability in enhancing sentiment analysis accuracy, trend detection, and user behaviour analysis, ultimately supporting informed decision-making for businesses across various sectors.