A modern production can involve hundreds of thousands of documents, and the privilege log is where the volume bites hardest. Privilege review is slow, expensive, and one of the most common flashpoints in discovery — a former magistrate judge has observed that logs that once ran fifty to a hundred entries now read like “little novels” of ten thousand entries or more, often at great cost and little use to anyone. The challenge at scale is to build a privilege log that is genuinely defensible without letting the logging exercise consume the case.
Why privilege review is a discovery bottleneck
Privilege is unusual in discovery because it requires a document-by-document legal judgment that cannot be fully automated. Every candidate has to be assessed against the elements of the attorney-client privilege or the work-product doctrine, and a wrong call in either direction carries a cost: over-withhold and you invite a challenge; under-describe and you invite a motion to compel. Multiply that judgment across a six-figure document set and the review becomes one of the slowest, most expensive phases of the matter. Because the log is produced near the end of discovery, problems with it tend to surface late, when deadlines are tight and patience is short.
What a privilege log must do
The Federal Rules do not actually use the phrase “privilege log.” The governing standard is Federal Rule of Civil Procedure 26(b)(5)(A): a party withholding privileged information must expressly make the claim and describe the nature of the withheld documents, communications, or things “in a manner that, without revealing information itself privileged or protected, will enable other parties to assess the claim.” That is the test for everything that follows. Whatever form the log takes, each entry must carry enough detail — the basis of the privilege asserted, and enough context to support the elements of attorney-client privilege or work-product protection — while disclosing nothing privileged on its face.
Logging methods at scale
There are three workable approaches, and the right choice depends on volume and complexity. The traditional document-by-document log gives a line-by-line description of each item — author, recipients, date, and a subject-matter description sufficient to justify the claim. It is the most thorough and the most burdensome, and it remains common in high-stakes matters. A categorical log groups similar documents into defined categories (for example, all communications between named counsel and the client over a date range) and assigns a common privilege description to the group. It is far more efficient at volume and is typically adopted by agreement or court approval; some jurisdictions actively promote it. A metadata or automated log generates entries from extracted fields — date, author, recipients, file type, title — without a narrative description, which is fast but leaves claims more exposed where context is needed to substantiate them. The trade-offs run along three axes: cost, defensibility, and the risk of either over-withholding or under-describing.
Clawback and inadvertent disclosure
At scale, some privileged material will slip through — the question is whether that produces a waiver. Two provisions of Federal Rule of Evidence 502 are the safety net. Rule 502(b) provides that an inadvertent production does not waive privilege if the holder took reasonable steps to prevent disclosure and acted promptly to rectify it. Rule 502(d) goes further: a court order can protect against waiver, and can preserve that protection even in other proceedings. The practical move is to negotiate a 502(d) order and a clear clawback provision in the ESI protocol at the outset, so a single misfired document does not become a waiver fight. Getting this in place early is one of the highest-leverage decisions in the whole review.
When privilege becomes a dispute
Even a careful log draws challenges. They tend to cluster around three things: the substance of individual designations, the adequacy of the log itself, and the review methodology that produced it. The receiving party often responds by demanding in-camera review — and at scale that demand collides with reality, because a court cannot realistically inspect thousands of documents one by one. The dispute then stalls, with the producing party defending its calls and the requesting party unable to test them.
How a neutral resolves it
This is where a special master or discovery neutral changes the dynamics. Rather than asking the court to shoulder an unmanageable in-camera burden, the parties can have the neutral review a sampled or contested set of entries, articulate a consistent standard for the close calls, and actually resolve them. A representative sample, decided against a single articulated standard, resolves far more than an entry-by-entry fight and gives both sides a predictable rule for the remainder. The neutral can also work at the front end — blessing a categorical-logging protocol before the log is built — so the format is agreed and defensible rather than litigated after the fact.
Practical takeaways
Three habits prevent most privilege-log disputes. First, agree the log format and a Rule 502(d) order early, in the ESI protocol, rather than after a problem appears. Second, use categories where the document population and the case allow it, so the log stays defensible without becoming a novel. Third, when contested designations outpace what counsel and the court can absorb, bring in a neutral to set a consistent standard and clear the backlog. The objective is the one Rule 26(b)(5)(A) demands: a log the other side can use to assess the claim, without revealing privileged information.
This article is provided for general informational purposes only and does not constitute legal advice. Engagement of Daniel Garrie as a neutral is administered exclusively through JAMS.