The documents that aren't there: discovery gap analysis
Every litigator knows some version of the 10 p.m. call: the deposition is tomorrow, the team has been reviewing the production for three weeks, and someone just realized there are no emails between the CEO and CFO for the one month that matters. Nobody missed a document. The document was never there — and three weeks of diligent review never surfaced its absence.
That’s the defining blind spot of document review: it examines what was produced. Nothing in a linear review workflow asks the more dangerous question — what should exist that doesn’t?
Absence is a finding
A production tells two stories. The first is in the documents’ contents. The second is in the production’s shape — and the shape is where the holes show:
- Custodian gaps. Six custodians were ordered; files from four arrived.
- Date-range gaps. Weekly board minutes with a five-week silence around the incident.
- Thread gaps. Email chains where replies reference messages that never appear.
- Category gaps. A trucking case with no driver-qualification file; an injury claim with no records from the treating specialist the plaintiff mentioned at deposition.
- Counterpart gaps. Your client’s copy of a correspondence exists; the other side’s side of it doesn’t.
Human reviewers are poorly built for this. Tracking completeness across tens of thousands of documents means holding a model of the expected universe in your head while reading the actual one — under deadline. It’s nobody’s failure that this breaks down; it’s an inventory problem wearing a reading problem’s clothes.
What gap analysis actually does
Gap analysis inverts the review. Instead of starting from the documents, it starts from a model of what a matter of this type, with these parties and this timeline, should contain — then audits the production against it:
- Inventory. Classify everything produced: type, custodian, date, Bates number, exhibit lineage.
- Expectation. Build the checklist of what should exist — from the case type, the pleadings, the discovery requests, and what the documents themselves reference.
- Difference. The gap list, ranked by how much each hole matters, with the evidence for why it should exist.
That last part is the point: a gap finding is only useful if it’s arguable. “You produced no maintenance logs for the vehicle for the six months before the crash, though your own inspection report references monthly service” is a meet-and-confer paragraph that writes itself.
In FactMarshal
This is built into the Materials surface. As the case file is ingested, every document is classified and tracked, and the coverage view keeps a running answer to “what do we have, and what’s missing?” — expected-but-absent records are ranked, each with the reasoning behind the expectation, and the follow-up request language is drafted from the gap itself. The 10 p.m. discovery happens at ingest, on week one, while there’s still time to do something about it.
Review tells you what the record says. Gap analysis tells you what the record is hiding by omission. You need both — because opposing counsel is only worried about one of them.