Published Paper Details:

Maintaining a healthy codebase over time is an ongoing hurdle for software teams. This study examines how different restructuring strategies perform when applied at key milestones--early development, major releases, and maintenance periods. We compare three approaches (manual refactoring, rule-based transformations, and AI-assisted restructuring) using four popular tools (RefactorX, Rearchitect, CodeRenew, SmartRefactor) under static, dynamic, and combined (hybrid) analysis. Across eight real-world projects and 2,592 experimental runs, we measured changes in code complexity, coupling, and code-smell counts, along with tool runtime and developer feedback. Our findings reveal that AI-powered methods deliver the largest complexity reductions (?25 %) with modest time investment, whereas rule-based tools strike a favorable balance between speed and effectiveness. Downtime around major releases offers the strongest opportunity for broad improvements, and hybrid analysis emerges as the most efficient way to catch both static and runtime issues. We close with clear, context-sensitive recommendations to aid teams in choosing the right restructuring strategy.