How Document Review at Scale Actually Works

Document review looks like a volume problem and behaves like a consistency problem. Ten reviewers applying a protocol slightly differently produce a production you cannot defend.

The protocol comes first

Before the first document is opened, the coding protocol is written down: what is responsive, what is confidential, what is potentially privileged, and the issue tags in play. Edge cases are decided in advance and recorded, so the same document gets the same call in week one and week four.

Batching and throughput

The population is divided into batches sized so that progress is measurable daily. A written staffing and throughput plan is agreed before the first batch, and reporting shows batches completed, documents coded, escalations raised and remaining volume against the deadline.

Quality sampling

A sample from every batch is re-reviewed independently. Disagreement rates are tracked; a rise usually means the protocol has met a fact pattern it did not anticipate, which is a protocol fix rather than a reviewer fix.

Escalation

Anything ambiguous — a document that may be privileged, a document outside the anticipated scope — is escalated to counsel with the reason, not resolved by a reviewer's instinct. Escalation volume is a health indicator, not a nuisance.

What stays with the firm

The final privilege determination, the privilege log as filed, the production decision and every representation to the court. We provide human review capacity inside your own review platform. We do not license or operate review software, we do not certify a production, and we do not make the privilege call.

Scope, staffing and reporting detail sit on the document review at scale page; the confidentiality controls applied to review teams are on security and confidentiality.