For a 10 TB time-series table in BigQuery, the best strategy is usually date or timestamp partitioning using the event date column, typically with daily partitions. Since most queries filter by date range, BigQuery scans only the required partitions instead of the entire table, greatly reducing cost and improving performance. To optimize further, you can add clustering on commonly filtered columns such as device ID, customer ID, or region. It’s also important to enforce partition filters to prevent accidental full-table scans and to set partition expiration policies for managing old data efficiently.