COUNT(DISTINCT col) in PostgreSQL silently disables parallel query execution because the aggregate has no combine function — workers can't merge partial distinct counts without shipping all values to one place. This forces a single-core serial sort that often spills to disk. The fix is to push deduplication into a GROUP BY subquery, which PostgreSQL can parallelize via partial hash aggregates. On a 10M-row table this rewrite yields ~3.4x speedup (1211ms vs 360ms) and eliminates disk spill. The same parallelism block affects any aggregate with an inner ORDER BY (string_agg, array_agg, percentile_cont). The per-group variant (GROUP BY country, COUNT(DISTINCT user_id)) has an analogous rewrite using stacked GROUP BYs, though the planner may not always choose the parallel path.
