Skip to main content
Use these best practices to improve privilege results, reduce rework, and keep output consistent across matters.

Plan the workflow before you run it

  • Define your privilege criteria before configuration begins.
  • Align on document types, labels, and expected log fields.
  • Confirm which metadata fields the workflow should rely on.
  • Decide how reviewers will validate and correct results.

Start with clean inputs

  • Use a dataset that matches the review scope.
  • Remove obvious duplicates before running the workflow.
  • Verify that documents contain usable text and metadata.
  • Confirm that participant, sender, recipient, and date fields are populated where possible.
Poor inputs lead to weak classification and incomplete log descriptions.

Prepare libraries early

  • Add known attorneys to the appropriate library before the first run.
  • Add known firms, clients, and organizations when available.
  • Use normalized names and remove duplicates.
  • Update libraries as new entities are identified during review.

Configure incrementally

  • Start with the default configuration when possible.
  • Change one area at a time.
  • Test each change before making additional adjustments.
  • Record what changed and why.
Small, controlled changes make it easier to trace issues and improve results.

Test on a sample first

  • Run the workflow on a pilot set before using the full dataset.
  • Check the output from each component.
  • Validate privilege calls on a representative mix of documents.
  • Review generated log descriptions for accuracy and consistency.
Use the sample run to catch gaps in metadata, instructions, or libraries early.

Review output at each stage

  • Do not wait until the end of the workflow to validate results.
  • Check entity labeling before relying on downstream classification.
  • Review potentially privileged documents first.
  • Inspect documents with missing or weak log descriptions.
Early review prevents errors from carrying into later workflow steps.

Use reviewer feedback to improve results

  • Capture repeated correction patterns from reviewers.
  • Refine instructions when the same issues appear across runs.
  • Update libraries with confirmed attorneys, firms, and domains.
  • Recheck edge cases after making changes.

Reuse what works

  • Save successful configurations for similar matters.
  • Apply proven instruction patterns across workflows where appropriate.
  • Compare results across matters to spot recurring issues.
  • Keep internal notes on validated approaches.