Sampling privilege logs for completeness, not privilege

The limited question

A privilege log can fail for administrative reasons even when the privilege is sound: a withheld document is missing from the log, a field is blank, an author is wrong, or a family is logged as a single document. This method tests those conditions. It does not assess whether a document is privileged, work product, or subject to waiver.

Define the comparison sets

Use the withheld and redacted sets as the population and the log as the record under test. Freeze the production date, the review date, and the log version. Record any documents added after the freeze rather than silently including them.

Set the field standard

The required fields come from the court’s order, the applicable rules, and the firm’s template. Typically they include a document identifier, date, author, recipients, type, subject, privilege asserted, and the basis approved by counsel. Write the standard down and attach the authority for each field.

Match every withheld item

For each withheld or redacted document, confirm a corresponding log entry. Then sample entries and verify the metadata against the document. Count unmatched items in both directions: withheld without a log entry, and log entry without a withheld document.

Review field accuracy

Score each sampled entry for missing fields, inaccurate metadata, and inconsistent descriptions. Preserve the first-pass result and adjudicate disagreements with the document open. Do not rewrite the privilege basis; return it to counsel with the document and the question.

Report and limit

Report the withheld count, the matched count, unmatched items, field accuracy, and the sample size. State clearly that the method tests administrative completeness and says nothing about the merits of any privilege claim.

Related research

• Sampling Bates productions for sequence and legibility errors
• Docket verification reliability: measuring alert-to-record agreement