Fixing BigQuery query structure cut a client's bill by 72% in three weeks

A data engineering team reduced a client's Google BigQuery costs by 72%, saving around £2,000 per month, in just three weeks by restructuring how queries were written rather than reducing data volume. BigQuery's on-demand pricing charges per byte scanned, meaning inefficient query design — not data size or traffic — is the primary cost driver. The biggest savings came from eliminating SELECT * statements, partitioning and clustering tables by commonly filtered columns, and enforcing partition filters to prevent accidental full-table scans. Several scheduled queries were also found to be feeding tables that no longer had any active readers and were paused and eventually deleted without any user complaints. The team noted that bytes scanned directly affects both cost and query latency, meaning structural fixes simultaneously reduced bills and improved dashboard performance.