Supporting the join
From text columns to a shared key.
One vendor, three strings
One vendor, three strings. An exact-match join drops two of them.
Read more →02 · ConnectFind the columns that describe the same dimension
Find the columns that describe the same dimension across datasets.
Read more →03 · JoinJoin any datasets, even without common keys
Sales and census population. Suppliers and company registries. One department’s figures and another’s. They rarely share an ID.
Read more →04 · OrganizeFiner levels roll up so sources meet at the right one
Finer levels roll up so sources meet at the right one.
Read more →05 · MaintainAmbiguous values go to a person
Ambiguous values go to a person. Decisions become aliases.
Read more →How it works
From raw values to monitored join keys.
CleanDims reads the schemas and distinct values of your sources, never whole rows. Every step a machine scores, a person approves, and each decision is kept and reused.
- 01Schemas and distinct values
No row data leaves the source. Upload files or connect systems.
- 02Columns grouped into dimensions
Five signals, one score. You approve each mapping.
- 03Unique values collated
Every value, with record counts.
- 04Values cleaned and grouped
One grouping per canonical value. At or above the threshold: auto-approved. Below it: you review.
- 05Canonical IDs published
Stable IDs, hierarchy levels and aliases for places, industries and vendors.
- 06Download or sync the outputs
Join any sources on the ID, at the right hierarchy level.
- 07Monitor
New values, new columns, drift. New values pass through the same review.
Scope
Where it fits.
CleanDims matches dimension values. Other tools match records or fields. It complements your warehouse, ETL, catalog and data-quality checks; it does not replace them.
| Approach | What it matches | What that means for a join |
|---|---|---|
| Entity resolution | Records about the same customer or company, in data you hold | Links rows, but gives two datasets no shared dimension |
| Master data management | Internal master data | Costly, and takes months to deploy |
| Integration mapping | Fields between systems | Does not touch the values inside those fields |
| SQL, scripts and LLM prompts | Whatever this project needs | Restarts on every project, with no shared canonical layer |
| CleanDims | Dimension values across open and internal datasets | One canonical layer with hierarchy levels, reused by every later join |
One canonical layer with hierarchy levels, reused by every later join.
Shared, reusable join keys
The outcome.
- Standardize
- One vendor, three strings. An exact-match join drops two of them.
- Connect
- Find the columns that describe the same dimension across datasets.
- Join
- Sales and census population. Suppliers and company registries. One department’s figures and another’s. They rarely share an ID.
- Organize
- Finer levels roll up so sources meet at the right one.
- Maintain
- Ambiguous values go to a person. Decisions become aliases.