Unitization

How does unitization work?

Electronic files usually arrive already unitized: a Word file is one document because the file system says so. The problem appears wherever that is not true.

Scanned paper is the classic case. A box becomes a stream of page images with no inherent boundaries. Someone must decide that pages 1 to 4 are a letter, page 5 is a fax cover, and pages 6 to 19 are the contract it transmitted. Done physically, the scanner follows the staples. Done logically, an operator or a classifier reads the pages and breaks where the content changes.

Chat and collaboration data is the modern case, and it is harder, because the data has no document shape at all. A Slack channel is a continuous log. Producing it requires an agreed unit, and the choice moves the document count by orders of magnitude from the same underlying data. The format risk planner generates protocol language fixing that unit.

Structured data has the same problem in a different form: a database has rows, not documents, and what gets produced is usually an agreed report or export rather than anything unitized at all.

Whatever is decided, the result is recorded mechanically. In the OPT file a Y in the document-break column on the first page of each document is the boundary, and everything downstream, Bates numbering, review batching, production ranges, follows from it.

What does unitization look like in practice?

A scanned batch, unitized two ways:

Physical and logical unitization compared on one scanned batch

Approach	Documents	What happened to the contract
Physical, follow the staples	41	Stapled behind its cover letter, so both are one document
Logical, read the content	63	Broken out as its own document with its own Bates range

Illustrative counts. The same 480 page images produce different document sets, different review units and different production ranges depending only on the boundary decision. Neither result is a processing error, which is why the approach belongs in the ESI protocol rather than in a vendor’s default settings.

How does unitization show up on an eDiscovery invoice?

Unitization is a scanning and imaging service on paper collections, priced per page or per document, and it is the step where manual work enters an otherwise automated pipeline. Logical unitization is billed higher than physical, or hourly, because a person is reading pages to find breaks.

On electronic data it is usually invisible until chat or mobile messaging appears in scope, at which point resolving the unit of production tends to arrive as project management hours rather than a per-unit rate.

Worth asking: is unitization included in the per-page scanning rate or billed separately, is it physical or logical by default, what is the rate for logical unitization, and who resolves ambiguous boundaries when the operator is unsure.

Sources

• EDRM, Processing Standards and Production Guide on document boundaries.
• The Sedona Conference, Glossary: eDiscovery & Digital Information Management, “unitization.”
• Fed. R. Civ. P. 34(b)(2)(E) on producing documents as kept in the usual course of business.
• Fee structure: the Fee Field Guide dataset.