You've written a query that returns the correct result but runs too slowly within the time window you have. Walk me through what you do, in order, from the moment you decide it's too slow.

When I notice a query is too slow, I first capture the exact query and the time window. I then run an EXPLAIN (ANALYZE) to see the execution plan and actual runtime statistics. I look for full table scans, missing indexes, or expensive joins. Next, I verify that statistics are up to date and that the optimizer has the right cardinality estimates. I check for data skew or large intermediate results. If necessary, I rewrite the query to use more selective predicates, push filters earlier, or replace subqueries with joins or CTEs. I also consider adding or tuning indexes, partitioning the table, or using materialized views. I monitor system resources (CPU, memory, I/O) to rule out contention. Finally, I test the changes in a staging environment, compare runtimes, and deploy once the performance meets the SLA. This systematic approach ensures I identify the root cause and apply the most effective fix.