The process of identifying, preserving, collecting, reviewing and producing electronically stored information in litigation or an investigation. Rule 34 of the Federal Rules of Civil Procedure is what makes it a discovery obligation rather than an IT exercise.

Who is responsible for eDiscovery in a case?

Counsel of record, whoever does the work. Obligations to preserve, to search reasonably and to produce in an agreed form sit with the party and its lawyers, and outsourcing the mechanics to a vendor does not move them. That is why a litigation hold is issued by the firm rather than by a provider.

Does the EDRM model describe what firms actually do?

Not literally, and it was never meant to. Stages overlap, repeat and get skipped, and several run at the same time rather than in sequence. Its value is as a shared vocabulary for talking about where a matter has got to, not as a workflow to follow step by step.

Which stages happen more than once in a matter?

Collection and production most often. New custodians surface, and productions are frequently rolling rather than single. Processing then repeats on each new tranche, which is why a per-gigabyte processing fee can be charged several times on one matter.

Where should someone new to eDiscovery start?

With the vocabulary and the bill, in that order. The glossary defines each term a protocol will use, and the Fee Field Guide decodes the line items those terms turn into. Between them they cover most of what an early conversation with a vendor will assume you know.

The eDiscovery process: nine EDRM stages and what each costs

The EDRM model names nine stages, from information governance through to presentation, and it is the vocabulary most vendors and courts use when they talk about where a matter has got to. It is a reference framework rather than a procedure: stages overlap, several repeat, and most matters never reach the last one at all. What follows is what happens at each, which fee category bills it, and where to read more.

What are the nine stages of the eDiscovery process?

The nine EDRM stages, what happens in each, and which fee category bills it
Stage	What happens	Billed as
Information governance	Retention policy and data hygiene, before any dispute	No discovery fee
Identification	Working out what data exists and who holds it	Services, usually project management
Preservation	Stopping routine deletion of what might be relevant	Services, sometimes per custodian
Collection	Taking data off the source without altering it	Services, per custodian or hourly
Processing	Turning raw containers into searchable documents	Processing, one-time per GB
Review	People reading documents and coding them	Review and access, per seat; hosting recurs throughout
Analysis	Grouping, threading and classifying to reduce what is read	Review and access, usually an analytics surcharge
Production	Delivering the responsive set in the agreed form	Production, per GB, per page or per production
Presentation	Using the material at deposition, hearing or trial	No discovery fee

Stage names are EDRM’s. The billing column maps each to a category in the Fee Field Guide, which decodes all nineteen fee types in full.

What happens before a single document is collected?

Three stages, and they are the ones a firm is most likely to skip and most likely to regret.

Information governance is everything a business does about its data before a dispute exists: retention schedules, what gets deleted and when. It carries no discovery fee because it is not discovery, but it decides how much there is to collect later.

Identification is working out what exists and who has it. In practice this is custodian interviews and a map of systems.

Preservation is the duty attaching. A litigation hold goes out, routine deletion stops for the people and systems in scope, and the clock on spoliation starts running. Getting this wrong is the one failure that cannot be corrected later by spending more.

Scope is argued at this end of the process too, which is where proportionality and the meet and confer belong.

Which stage does the money actually go to?

Review, by a wide margin.

The RAND Institute for Civil Justice study Where the Money Goes found review accounting for roughly 73 percent of production cost, processing roughly 19 percent, and collection roughly 8 percent. That distribution is the single most useful thing to know about the process, because it explains why every meaningful cost lever sits upstream of the stage that spends the money.

The stages before review exist largely to make review smaller. Culling removes by rule, de-NISTing drops known system files, and deduplication collapses identical documents. Everything those remove is a document nobody pays to read.

The billing works in the opposite direction from the effort, which is the structural oddity of the whole model. Processing is charged once, per gigabyte. Hosting is charged every month, on everything retained, whether or not it is ever opened. So the cheapest possible matter is one where the volume was cut before the meters started.

Does every matter run every stage?

No, and a page that implied otherwise would be misleading.

Presentation is the clearest case: most matters settle, so the material is never used at a hearing. Analysis is routinely skipped on collections small enough to read linearly, since technology-assisted review earns its keep at volumes a small team cannot reach. Information governance may not have happened at all, which is usually discovered during identification.

Others repeat rather than skip. Collection runs again when a new custodian surfaces. Document production is frequently rolling, in several volumes on a schedule, and each one carries its own production fee.

Treating the nine as a sequence to work through is the most common misreading of the model. They are a map of the territory, not a route through it.

Where does the process most often go wrong?

At the handovers, rather than inside any single stage.

Between preservation and collection, when a hold is issued but never followed up, so the data it was meant to protect is gone by the time anyone collects.

Between collection and processing, when what was collected does not match what was scoped, and the gap surfaces as a chain of custody question rather than a technical one.

Between review and production, which is where most visible failures happen: a production that fails to load, a privilege log that does not match what was withheld, a redaction applied to an image but not to the native underneath. The production checklist exists for this handover specifically.

The pattern is that each stage is usually performed competently by someone who can see only their own stage.

How this works on a flat rate

Kavarn, the flat-rate eDiscovery service Konvergence Inc operates, covers processing, hosting, review access and production in one instance rate rather than metering each stage separately. The stages still happen in the same order and the same work still gets done; what changes is that no stage carries a separate charge, so the decision about whether to collect a further custodian is about the matter rather than about the invoice.

That is a narrower claim than it sounds. It does not make review cheaper, because review is people reading documents and that cost sits with the firm either way. What it removes is the per-gigabyte charge that attaches to holding the data while they do. The pricing page publishes the rate, and the cost calculator compares the two models on your own figures. See what the platform includes, or book a walkthrough.

Sources

EDRM for the stage model and its names.
RAND Institute for Civil Justice, Where the Money Goes: Understanding Litigant Expenditures for Producing Electronic Discovery (Pace & Zakaras, 2012), for the distribution of cost across review, processing and collection.
Fed. R. Civ. P. 34 on the obligation to produce ESI, and Fed. R. Civ. P. 37(e) on failure to preserve.
Fee categories for each stage: the Fee Field Guide dataset.