QueryFuser
Reduce BigQuery bytes scanned with query merging
QueryFuser is a PostgreSQL-compatible proxy that sits between your BI tools and Google BigQuery. Your dashboards, reports, and ad-hoc queries connect to QueryFuser exactly like they would to any PostgreSQL database — no code changes, no SDKs, no plugins.
What it does:
When a dashboard loads, BI tools like Looker, Tableau, Metabase, and Power BI fire a separate query for every chart tile. Each query scans the same underlying tables independently, and BigQuery bills you for every scan. QueryFuser intercepts these concurrent queries, detects that they hit the same tables, and merges them into a single BigQuery request. The data is scanned once, and each caller gets back exactly the rows they asked for. On top of merging, QueryFuser applies cache optimization — rewriting volatile expressions like CURRENT_DATE() into stable literals so BigQuery's built-in cache can serve repeated queries at zero cost.
Who it's for:
Teams running BI dashboards on BigQuery's on-demand pricing. If you have analysts, product managers, or execs opening dashboards throughout the day, you're paying for the same table scans over and over. QueryFuser is designed for companies spending $1K–$100K+/month on BigQuery who want to cut that bill without rearchitecting their data stack.
Why we built it:
BigQuery charges per byte scanned, not per question answered. An 8-tile dashboard scanning a large fact table can cost 8× what it should. We saw teams trying to solve this with materialized views, BI-level caching, or flat slot reservations — all of which add complexity or shift costs. QueryFuser solves it at the wire protocol level: your tools think they're talking to PostgreSQL, and QueryFuser handles the rest. One scan, many answers, lower bill.
What it does:
When a dashboard loads, BI tools like Looker, Tableau, Metabase, and Power BI fire a separate query for every chart tile. Each query scans the same underlying tables independently, and BigQuery bills you for every scan. QueryFuser intercepts these concurrent queries, detects that they hit the same tables, and merges them into a single BigQuery request. The data is scanned once, and each caller gets back exactly the rows they asked for. On top of merging, QueryFuser applies cache optimization — rewriting volatile expressions like CURRENT_DATE() into stable literals so BigQuery's built-in cache can serve repeated queries at zero cost.
Who it's for:
Teams running BI dashboards on BigQuery's on-demand pricing. If you have analysts, product managers, or execs opening dashboards throughout the day, you're paying for the same table scans over and over. QueryFuser is designed for companies spending $1K–$100K+/month on BigQuery who want to cut that bill without rearchitecting their data stack.
Why we built it:
BigQuery charges per byte scanned, not per question answered. An 8-tile dashboard scanning a large fact table can cost 8× what it should. We saw teams trying to solve this with materialized views, BI-level caching, or flat slot reservations — all of which add complexity or shift costs. QueryFuser solves it at the wire protocol level: your tools think they're talking to PostgreSQL, and QueryFuser handles the rest. One scan, many answers, lower bill.