Would you validate that changes improved performance? Would you check EXPLAIN ANALYZE?

To validate performance improvements, first capture baseline metrics: execution time, I/O, CPU, and memory usage using EXPLAIN ANALYZE or a profiling tool. After making changes (e.g., adding indexes, rewriting queries), run the same query and compare the plans. Look for reductions in sequential scans, smaller row counts, and lower cost estimates. Use pg_stat_statements or equivalent to aggregate query performance over time. If the plan shows a different join algorithm or fewer rows processed, and the execution time is lower, you can conclude the change helped. Always test against representative data volumes and consider edge cases to ensure the improvement holds under load.