performance-optimization

Fix N+1 queries by replacing per-record lookups with a single query using a join or include clause, such as findMany with include in an ORM. Confirm the fix by checking the database query log before and after the change.

Improve Core Web Vitals by first measuring with Lighthouse and the web-vitals library, then targeting the failing metric: optimize hero images and render-blocking resources for LCP, reduce main-thread long tasks for INP, and set explicit image dimensions for CLS.

Add an index only after running EXPLAIN ANALYZE and reading the query plan. Index for the shape of the query with equality columns first, and skip indexing for low-selectivity filters, leading wildcards, or write-heavy tables where the write cost outweighs read gains.

Simultaneous slowness across all endpoints is the signature of connection pool exhaustion, where requests wait for connections rather than execute. Use one pool per process sized against the database limit, and add a proxy like pgbouncer for serverless or autoscaling environments.

Cache only data that is expensive to produce and read far more often than it changes. Include every input that affects the response in the cache key, pick one invalidation strategy such as TTL or versioned keys, and never cache data where staleness is a correctness bug.

Do not optimize before you have profiling evidence of a real problem. Premature optimization adds complexity that costs more than the performance it gains, and changes that show no measurable improvement over baseline should be reverted rather than kept.