Find the same customer under three different codes.
Record Matching compares records by rules you set, queues the uncertain pairs for a person to review, and exports decisions, a crosswalk and a deduplicated file.
Two systems, or one old one, hold the same customer under different names, codes and spellings. Merging or migrating without matching them first doubles your customer list.
Modules. Record Matching works without a paid module up to 1,000 rows per file. Transfer lifts that limit.
Edition. DataBridge Server.
How DataBridge does it.
Each step names the feature as it appears in the product.
- Step 1
Load files
Drop one CSV or TSV to find duplicates inside it, or two to match across systems. Source files are never changed.
- Step 2
Rules
Compare by Exact, Normalized (case, punctuation, legal suffixes such as Inc and LLC, street abbreviations, digits only) or Fuzzy with a minimum similarity. Roles include Weighted, Must match, Must not match and Decisive, so an exact tax ID can beat a weak name score.
- Step 3
Review
Pairs above the auto-match threshold are pre-marked, pairs between the two thresholds go to a review list with keyboard shortcuts, and lower pairs are dropped.
- Step 4
Export
Export Decisions, a Crosswalk (old code, new code, status Confirmed, Proposed or Conflict) and, for one file, the Survivors file with absorbed codes in _MergedKeys.
- Step 5
Use the crosswalk
Import it into Crosswalks and apply it in field mappings, so every later load translates codes the same way.
What this does not do.
- A value shared by thousands of records is too common to compare, and the result says so. Add a more selective rule, such as a tax ID or postal code.
- DataBridge proposes and records decisions. People confirm them.
Related pages.
Try the parts that are free today.
The free browser edition needs no sign-in and never uploads your files.