Skip to main content

Configure standardization rules

Before you start​

  • The attribute already defined.
  • The standardization-rule.create permission.

1. Choose the standardizer​

The new standardization rule form A new standardization rule. The standardizer chosen here decides which options the form then offers.

Match it to the attribute's meaning, not merely its data type. An email column and a free-text note are both strings; only one benefits from email normalization.

2. Set execution order​

Standardization rules ordered per attribute Rules on one attribute, in execution order — each receives the previous one's output.

Rules on an attribute run in a defined order, and the order changes the result. Stripping separators and then uppercasing produces a different value from uppercasing and then stripping.

Order them deliberately: normalize shape first, then case.

3. Save — and expect it to be refused if wrong​

A rule whose configuration the engine cannot honour is rejected when you save, not accepted and skipped at ingest.

This is the behaviour you want. A silently ignored rule produces data that looks standardized, matches badly, and gives no signal about why. Failing at configuration time makes the problem visible while you are still looking at it.

4. Test before loading​

Run a value through and inspect the result. Test the awkward cases, not the easy ones:

  • A value already in canonical form — it should be unchanged, not double-processed.
  • An empty or whitespace-only value.
  • A value in an unexpected format.
  • For addresses, one that is genuinely messy.

5. Apply to existing records​

Standardization runs on write. A rule added after loading does not retroactively rewrite what is already stored.

To apply it to existing records, run a recompute — see load data in bulk.

Standardize before matching, always

Matching compares canonical values. Matching data that predates a standardization rule compares inconsistent forms and produces false negatives — records that should match and do not. If you add a rule after loading, recompute before you match.

Next​

Last verified against commit a0765982 (2026-08-03)