You are given an existing program that produces correct results but is too slow in production. Describe a structured plan to optimize it. Your answer should explain how you would define success metrics, collect baseline measurements, profile the program, identify the real bottleneck, decide whether the issue is algorithmic, memory-related, I/O-bound, synchronization-bound, or compiler-related, and then choose optimizations such as better algorithms, improved data layout, batching, caching, vectorization, parallelism, and compiler settings. Also explain how you would verify correctness and prevent performance regressions.