BigQuery incentivizes efficient data retention by cutting storage costs in half for tables or partitions that go 90 days without modification. However, many teams unintentionally forfeit this discount by performing broad or unnecessary updates to long-lived datasets — for example, touching an entire table when only a few rows need to change. Even small-scale or programmatic updates can trigger a full reset of the 90-day timer on affected data. This behavior is subtle but expensive: it silently doubles the storage cost of large datasets for another 90-day cycle, even when the data itself is mostly static. Without intentional safeguards, organizations may repeatedly reset their discounted storage window without realizing it.
BigQuery storage is billed based on: