What is Document Review?
Document review is the process of examining a body of documents to assess their relevance, privilege status, and significance to a legal matter. You see it most often in litigation discovery, mergers and acquisitions due diligence, regulatory investigations, and contract analysis, where a party must read, classify, and code large volumes of material before producing or relying on it.
The goal is to decide three things about each document: whether it is responsive to the issue at hand, whether it is protected from disclosure (for example by privilege), and what facts or risks it reveals. Reviewers tag or "code" documents along these lines so that the relevant ones can be produced, withheld, or escalated.
What document review involves
A typical review breaks into a few core decisions made on each document:
- Relevance (responsiveness): Does the document relate to the claims, defenses, or transaction at issue, and does it fall within the scope of a discovery request or due diligence checklist?
- Privilege: Is the document shielded by attorney-client privilege or the work-product doctrine, meaning it should be withheld or produced on a privilege log rather than disclosed?
- Key facts and issues: Does the document contain hot facts, admissions, deal risks, or material terms that the legal team needs to flag?
- Confidentiality and redaction: Does it contain trade secrets, personal data, or other sensitive content that must be redacted or designated under a protective order?
In litigation, document review sits at the heart of the discovery phase and modern e-discovery, where most relevant documents now exist as electronically stored information (ESI). In transactional work, the same process drives due diligence and contract review.
Manual vs technology-assisted review
There are two broad approaches, and most large reviews now blend them.
Manual (linear) review means human reviewers, often associates or contract attorneys, read documents one by one and code each by hand. It offers nuanced judgment but becomes slow and expensive at scale.
Technology-assisted review (TAR), also called predictive coding, uses machine learning to prioritize or classify documents. Subject-matter experts code a sample set, and the algorithm extends those decisions across the larger population, ranking documents by likely relevance.
| Aspect | Manual review | Technology-assisted review (TAR) |
|---|---|---|
| Method | Human reads each document | Algorithm trained on coded samples |
| Speed at scale | Slow | Faster on large volumes |
| Typical cost | Higher per document | Lower per document at scale |
| Consistency | Varies by reviewer | More consistent classification |
| Court acceptance | Long-established | Approved in U.S. federal practice |
In the United States, courts have accepted TAR for years. The 2012 federal decision in Da Silva Moore v. Publicis Groupe is widely cited as the first to approve predictive coding for document review, and later rulings reinforced that producing parties may generally choose TAR. Acceptance and specific protocols vary by jurisdiction and by judge, so you should confirm what is expected in your forum.
Privilege screening
Privilege screening is the part of review where you identify and withhold documents protected by privilege or work product. Getting this wrong has real consequences: producing a privileged document to the other side can, in some circumstances, waive the protection.
In U.S. federal litigation, parties often use a clawback arrangement under Federal Rule of Evidence 502(d) to recover inadvertently produced privileged material, and Federal Rule of Civil Procedure 26(b)(5)(B) sets out the notify-and-sequester steps after an inadvertent disclosure. These are jurisdiction-specific safeguards, and a clawback order does not excuse a careless review, so reviewers still apply privilege coding carefully and maintain a privilege log.
Where document review applies
- Litigation discovery: Reviewing collected ESI and paper records before producing documents to an opposing party or responding to a subpoena.
- M&A due diligence: Examining contracts, corporate records, and compliance files to surface liabilities and material terms before a deal closes.
- Regulatory and internal investigations: Reviewing communications to assess exposure and respond to government requests.
- Contract analysis: Reading agreements to extract obligations, renewal dates, and risk provisions across a portfolio.
Why document review matters
Document review is often the single most expensive and time-consuming part of litigation and large transactions, and the quality of the review shapes the outcome. A thorough review surfaces the facts that win or settle a case and protects privilege; a sloppy one can leak privileged material, miss a smoking-gun document, or produce material outside the proper scope.
It is also where accuracy and speed collide, because reviewers must process enormous volumes without sacrificing judgment on close calls. That is why teams increasingly pair human expertise with software, from TAR platforms to AI-assisted classification, to triage documents faster while keeping lawyers in control of the final calls. An AI legal assistant like LegesGPT can help you surface relevant passages, flag potential privilege, and summarize key facts across a document set, so you spend your time on the judgment calls that actually require a lawyer.
Frequently asked questions
What is document review in litigation?
Document review in litigation is the process of examining collected documents during discovery to decide whether each is relevant to the case, protected by privilege, or contains key facts. Reviewers code documents so the legal team can produce, withhold, or flag them. It is usually the most time-consuming and costly part of discovery, especially with large volumes of electronically stored information.
What is the difference between manual review and technology-assisted review (TAR)?
Manual review means human reviewers read and code each document by hand, which gives nuanced judgment but is slow and expensive at scale. Technology-assisted review, also called predictive coding, trains a machine-learning model on documents coded by experts and applies those decisions across a larger set, making review faster and more consistent on big volumes. U.S. courts have accepted TAR since the 2012 Da Silva Moore decision, though protocols vary by jurisdiction.
What is privilege screening in document review?
Privilege screening is the step where reviewers identify documents protected by attorney-client privilege or the work-product doctrine and withhold them from production, logging them on a privilege log. In U.S. federal litigation, clawback orders under Federal Rule of Evidence 502(d) and the procedures in Rule 26(b)(5)(B) help recover privileged material that is produced inadvertently, but rules and outcomes vary by jurisdiction.