Table statistics are essential for optimal query plans

The query planner looks at available indexes, cost settings, strategy settings, and distribution of data, in order to plan the most efficient way to execute a query.

Stats about the distribution of data are stored in the pg_statistic table and presented in a more readable form in the pg_stats view. They are generated based on ~20% of the table data (less for large tables) and are constantly updated by the autovacuum daemon. You can also manually update them using ANALYZE or VACUUM ANALYZE, the later which is strongly recommended after inserting or deleting a large number of rows.