7 BigQuery Cost Optimisation Tips That Actually Work

The BigQuery Bill Shock

BigQuery's on-demand pricing charges $5 per TB scanned. That sounds cheap until your analysts are running exploratory queries against a 10 TB table multiple times a day. I've walked into organisations paying $30k/month on BigQuery that got it down to under $8k with the changes below.

1. Partition Your Tables

Partition by date (ingestion time or a date column). A query with a WHERE date >= '2024-01-01' filter will only scan the relevant partitions — often reducing data scanned by 90%+.

2. Cluster on High-Cardinality Filter Columns

After partitioning, add clustering on columns you frequently filter on (e.g., user_id, country). BigQuery will physically co-locate matching rows, reducing scan size further.

3. Use Materialised Views for Hot Aggregations

If the same aggregation is queried dozens of times per day, create a materialised view. BigQuery auto-refreshes it incrementally and queries hit the pre-computed result instead of the raw table.

4. Prefer Slots + Reservations for Predictable Workloads

If you're spending more than ~$2,500/month on on-demand, evaluate BigQuery Reservations (flat-rate). 100 baseline slots at the Standard edition currently costs roughly $2,000/month — often cheaper than on-demand at scale.

5. SELECT Only What You Need

BigQuery is columnar — SELECT * scans every column. Always project only the columns you need. This one change alone can cut costs 30–50% in wide tables.

6. Use BI Engine for Dashboard Queries

Looker Studio and other BI tools hammer the same queries repeatedly. BI Engine caches results in-memory for a fixed hourly fee, drastically reducing on-demand charges from dashboards.

7. Set Cost Controls

Set per-query byte limits (maximumBytesBilled) so rogue exploratory queries can't run up a surprise bill. Combine with dataset-level custom quotas per user.

Implementing all seven can realistically reduce a $20k/month BigQuery bill to under $8k. Get in touch if you'd like a BigQuery cost audit.