Redescription Mining,
in Real SQL

Spile reads the tables already in your warehouse and looks for pairs of unrelated queries that return nearly the same rows. Two descriptions of one group, built from completely separate columns, that nobody thought to compare. You get each finding as SQL you can run yourself, alongside a prose description of the mined pattern.

Your data never moves. Spile runs inside your own cloud project under a service account you grant and revoke, so access is a permissions question rather than a file transfer. Findings that match only a handful of rows are dropped automatically — a query matching three rows identifies three people — and that limit is structurally enforced in the miner.

Upload your data

{{ dropHint }}

CSV — up to 50 MB

{{ fileName }}

{{ fileSize }} · ready

remove

{{ checkoutLabel }}

Analysis job started

{{ jobNote }}

Could not start the job

{{ errorMessage }}

Job ID

{{ jobId }}

Free Trial {{ footNote }}