Forms, reference tables and CSV mappings determine what a label decision receives. Author edits these drafts; Reviewer reads the saved work and its frozen evidence before publication. A correct-looking layout does not prove that the incoming values or lookup result are right.
Prepare the form definition
Open Operator forms and choose the intended draft. Check each field identifier, label, type, required status, choices and visibility conditions. Under Permitted corrections, choose only the supported correction catalogue and intended bindings. A prompt-only field does not change imported job data.
Use Add field and the individual field controls deliberately. Review dependencies before removing or changing a field referred to by another condition. A hidden required field can hold a trial rather than quietly satisfying its requirement. Keep the form reason and review acknowledgement meaningful, then choose Save new revision.
Use Try saved revision for transient feedback. Trials are not retained independent examples and create no label. Follow whole form-workflow proof for form-only examples, complete source outcomes and the combined candidate. Production Operators use the published definition, not this draft or trial.
Review and edit a reference table
Open Data and select the correct table draft. Confirm its declared table type and exact matching keys before changing rows. Edit the exposed text, value and date fields; use the shown add/remove controls only for the intended rows.
Review exact identifiers, values and effective dates. Preserve leading zeros and meaningful spaces. Duplicate identities or overlapping effective periods for the same matching key need correction against the actual requirement, not a guessed new code. An empty optional end date has its stated open-ended meaning.
Check the acknowledgement and review note, then choose Save reviewed table. Confirm the new saved revision before testing it. Discard or save unsaved table edits before running a saved-source check.
Import a CSV table through a retained review
From the intended table, open Import table from CSV. Choose pasted CSV or a file and use the exact headers shown for that table, in the offered comma-separated UTF-8 format. Keep dates in the stated YYYY-MM-DD form and inspect the displayed row/file limits.
Choose Merge by exact row ID to update matching rows and append new rows, or Replace the complete table to remove retained rows absent from the CSV. A permitted header-only replacement can remove all preferences. It is not equivalent to importing no change.
Validate the CSV and inspect Retained import review, including exact additions, changes and removals. Validation creates a proposal, not a new draft revision. Resolve stale-revision or blocked-review messages before relying on it.
Under Save the reviewed import, acknowledge the exact keys, values, dates and changes. Confirm removals separately when listed and provide Import review note, then choose Save as a new draft revision. Return to the draft and inspect the actual saved values. A successful import still needs source-example proof and separate publication.
Check the lookup against independent expectations
Use Check a raw source example on the saved table. Supply the actual example context and independently expected result, including the expected exact match or No matching row (null). Check missing and ambiguous conditions as well as the normal match.
Inspect the lookup result and Proposed preference and pinned table. Keep the table revision and retained check with the intended release. A parsed CSV alone does not prove the receiving lookup or final label is right.
Build the CSV mapping in order
Open the mapping draft and confirm the source format and target field under Column mapping. Choose the exact source CSV header for each declared target. A header name and a target identifier are different values.
Review Ordered transformations from first to last. Where offered, string operations include trimming, ASCII case conversion, literal prefix/suffix and literal replacement. Date targets can use their declared source-date parser. Changing operation order can change the result, so use the move/remove controls deliberately and test the exact sequence.
Without a transformation, the source value is preserved and then checked against its declared type. Do not assume quantities, dates or identifiers accept whatever a spreadsheet happened to export. Save the reviewed mapping with its acknowledgement and note, and confirm the current saved revision.
Verify a raw row and retain its revision
Use Check one raw CSV row after saving or discarding edits. Supply the raw source and independently expected field values. Inspect the saved mapping result, diagnostics and downstream values. An ordinary parse is not proof that a business field was mapped correctly.
Under Revision history and export, retain the required revision where appropriate. Mapping JSON exports preserve exact strings and transformation order; importing that exported mapping format is not currently supported. Do not promise an import/restore operation because an export exists.
Freeze and hand over
Each editor's checks belong to its exact saved revision. Changed form, table or mapping content needs current evidence and its applicable candidate workflow. Reviewer checks the frozen input configuration and output expectations through verification and publication.
Data delivered by another SmartFACT product still needs the receiving LabelRules review. Delivery alone is not approval. Keep source filename/row, exact value, draft/revision, retained check and expected outcome in a private handover; share only necessary records and never public customer source files.