QueryFuser

Teams running BI dashboards on BigQuery pay for every byte scanned on every refresh. When a dashboard has multiple tiles, most BI tools fire one independent query per tile even when those queries hit the same underlying tables, so the same columns get scanned repeatedly. An 8-tile dashboard on a star schema can run 16 column-scans worth of work when only 4 are actually needed. Because BigQuery bills per byte scanned and consumes slot time per query, this redundant querying directly inflates the bill and eats into slot capacity on every single dashboard load.

QueryFuser sits between BI tools and BigQuery as a transparent proxy that speaks the PostgreSQL wire protocol, so any tool that can connect to Postgres (Looker, Looker Studio, Metabase, Grafana, Superset, Redash, Tableau, Sigma, Rundown BI, or generic clients like psql, DBeaver, JDBC/ODBC) can point at it with standard host, port, user, and password settings. It intercepts incoming queries in Postgres dialect, translates them automatically to BigQuery SQL, and detects when multiple queries from the same dashboard load hit the same tables, fusing them into a single BigQuery query before splitting the results back to each caller. Additional features include an Auto Cache Optimizer that replaces non-deterministic functions like NOW() and CURRENT_DATE() with literals so repeated dashboard loads can hit BigQuery's 24-hour result cache, and centralized access control that removes the need to distribute GCP service account keys or IAM access to every user and tool. The site claims 35 to 85 percent reductions in bytes scanned depending on schema, join key types, and column overlap, with worked examples showing 75 percent savings for INT64 star schemas, 68 percent for STRING key star schema