The rapid growth of social media platforms has heightened competition among marketers to allocate advertising budgets effectively while maximizing return on investment (ROI). Predictive analytics, leveraging advanced machine learning algorithms and data-driven forecasting techniques, has emerged as a promising approach to improve ad spend efficiency through enhanced targeting, bid optimization, and performance prediction. This study systematically evaluates the effect of predictive analytics on social media advertising outcomes by combining quantitative modeling with empirical analysis across multiple platforms, including Facebook, Instagram, and LinkedIn, over a 12-month period. Employing regression, time-series forecasting, and ensemble learning methods, we predict consumer engagement and conversion rates, and compare predictive-model-driven campaigns with conventional heuristic budgeting strategies. Findings indicate that predictive analytics improve cost efficiency by up to 28%, with significant gains in click-through and conversion rates. Notably, improvements vary across platforms and industries. The study highlights predictive analytics as a critical strategic tool while acknowledging limitations regarding data quality, algorithmic bias, and platform-specific factors. Contributions include empirical benchmarks and actionable recommendations for practitioners seeking to optimize advertising budgets in an increasingly competitive digital landscape.