A scheduled query is a managed recurring job, configured outside the SQL worksheet. Its SQL is only one part of the work. You also choose the destination, schedule, identity and location, then monitor successful runs and late or failed data. This console doesn't contact Google Cloud or start a scheduler. You can run the query logic locally and make the operational choices in checked scenarios.
Scheduled queries support @run_time and @run_date parameters. A backfill should use the intended run boundary rather than whichever date happens to be current when a retry executes. Decide what a repeated run does before enabling the schedule: replacing one reporting partition or merging on a stable key can be repeatable, while an unchecked append can duplicate results. Daylight saving time, late source arrivals and revoked credentials all deserve an explicit operating rule.