Why a Seq Scan instead of an Index Scan?

The planner compares the estimated cost of plans based on statistics and the query's conditions. A sequential read can be better for a small table or a query that returns a large share of the data.

EXPLAIN (ANALYZE, BUFFERS)
SELECT id, created_at
FROM orders
WHERE customer_id = 42;

I check: estimated versus actual row counts, the number of loops, buffer reads, sorting and timing. A large estimation error can point to a statistics problem.

Pitfall: EXPLAIN ANALYZE actually runs the statement. For writes it changes data — it isn't a harmless preview. A Seq Scan alone doesn't prove there's a problem.