"eDiscovery consultant" means four different things depending on who is saying it, and the ambiguity costs people real money. A general counsel who believes she is buying strategic advice and receives a hosting invoice has bought the wrong thing. So has the one who hires a vendor and then expects that vendor to defend the methodology in a deposition.

What follows is the honest version of the role: what it covers, what it does not, when it earns its fee, and when you should keep the money.

Four eDiscovery roles that get called the same thing

Start by separating the categories, because confusion between them is the source of most disappointing engagements.

  • The consultant advises on process and defensibility. The output is decisions and documentation: scope, methodology, protocols, workflows, and a record showing the choices were reasonable. A consultant is accountable for whether the approach holds up under scrutiny.
  • The service provider executes. Collection, processing, hosting, managed review, production. The output is data, moved and transformed. Providers are accountable for delivering against a specification. They are generally not accountable for whether the specification was the right one.
  • The platform is software. Relativity, Everlaw, and their competitors are tools, and a tool has no opinion about your custodian list.
  • The testifying expert offers an opinion under oath, subject to disclosure obligations, Daubert scrutiny, and cross-examination. That role carries independence requirements the others do not, and the testifying expert is not automatically the person who did the consulting.

Plenty of firms do more than one of these, which is fine in itself. It does change the incentives, and you should know which hat is on at any given moment. A provider who also advises on collection scope has a structural reason to prefer a larger collection. That does not make the advice wrong. It makes it worth asking about.

The distinction that matters

A vendor answers the question you asked. A consultant tells you when you asked the wrong question. If nobody on the matter is doing the second job, your outside counsel is doing it by default, whether or not counsel has the background for it.

What the work actually looks like

Before anything is filed

Readiness work: data mapping, retention alignment, hold-process design, and building the playbook that makes the next matter cheaper than the last. This is the highest-return phase and the one organizations most often skip, for the simple reason that no deadline forces it.

At the trigger

Scoping the preservation duty. Which systems, which custodians, which date range, and which of them can be defensibly excluded. The consultant's contribution here is frequently subtractive: talking a team out of preserving everything, on a documented rationale that will survive being read back to them two years later. Over-preservation is not a safe default. It is a deferred cost carrying its own risks.

Meet and confer

Drafting and pressure-testing the ESI protocol. Formats, metadata fields, search methodology, validation, and how modern collaboration data will be rendered. The Rule 26(f) conference sets the cost of everything downstream, and it is routinely staffed by people who have never had to produce a Slack workspace.

Collection and processing

Method selection for each source, vendor oversight, and validation that what came back is what was supposed to come back. Counting documents against expectations. Noticing the mailbox that returned suspiciously small.

Review

Search-term negotiation and testing, workflow design, and, where technology-assisted review is used, the validation protocol and the documentation that makes the result defensible. The technology is not the hard part. The record of how you validated it is.

When something has gone wrong

Gap analysis, remediation planning, declarations, and reconstructing what happened and when. This is the most expensive way to buy a consultant, and it is how most people first buy one.

The moments that justify calling an eDiscovery consultant

  1. A preservation notice whose scope you cannot confidently draw. Unfamiliar systems, cross-border data, ephemeral messaging, or a custodian population nobody can enumerate. The cost of getting this wrong compounds silently for months before anyone sees it.
  2. A Rule 26(f) conference on data sources you do not know. If the matter turns on Teams, Slack, a bespoke line-of-business database, or product telemetry, you are about to negotiate a protocol whose consequences you cannot price.
  3. An opposing ESI protocol you cannot evaluate. Most protocols contain one or two clauses that quietly determine the budget. If you cannot identify which ones, do not sign it.
  4. A sanctions motion, or a credible threat of one. Rule 37(e) turns on reasonableness and, for the severe measures, on intent. Both are proven with a contemporaneous record that someone has to assemble and then explain.
  5. An investigation or M&A matter with cross-border data. GDPR, national blocking statutes, and works-council obligations interact with US discovery in ways that are neither intuitive nor forgiving. The data map is the starting point.
  6. A production whose methodology will be attacked. Not the documents, the method. When that is the likely fight, the documentation has to be built before the fight, not after.

When you do not need one

Most firms leave this part out. Here it is plainly.

  • Small, single-custodian matters on ordinary systems. One mailbox, one OneDrive, a clean date range, a familiar platform. Competent outside counsel and a decent provider can run this. A consultant adds coordination, not defensibility.
  • Repeat matters you already have a process for. If you handle forty similar employment cases a year and the workflow is documented and working, the consulting was done when the workflow was written. Do not re-buy it annually.
  • When what you need is capacity. If the real problem is that nobody has time to run the project, hire a project manager or a managed-services provider. That is a staffing problem with a staffing solution, and it costs less.
  • When outside counsel genuinely has the depth. Some litigation groups have real eDiscovery bench strength, with people who have argued these issues rather than delegated them. Ask directly. A firm that has it will be glad to tell you, and a firm that does not will usually say so rather than improvise.
  • Retrospectively, on a closed matter nobody challenged. A post-mortem on a production that drew no objection is an expensive way to feel reassured. Spend the money on readiness for the next matter instead.

How to evaluate one

The standard sales conversation is about tools and throughput: platforms certified, terabytes processed, matters supported. None of that predicts whether an engagement will be useful to you.

Ask instead what you will be holding when it ends.

  • What are the deliverables, by name? A written scope memo. A draft ESI protocol. A validation report. A documented workflow. A declaration someone is willing to sign. If the answer is "ongoing advice," you have bought an expense rather than an asset.
  • Who does the work? The named person in the pitch, or somebody else entirely. Get the answer into the engagement letter.
  • Have you testified? Not because every matter needs a testifying expert, but because someone who has been cross-examined on their own methodology writes documentation differently, and permanently.
  • Where is your independence? If they also sell hosting and review, ask how the recommendation is insulated from the sale. A good answer exists. Insist on hearing it rather than assuming it.
  • What would you tell me not to do? The most useful question in the room. Anyone who cannot name something you should skip is selling scope.

One more, and it separates practitioners from brochures: ask them to describe a matter where their advice turned out to be wrong, and what they changed afterwards. Everyone who has done this work for a decade has an answer. Only some of them will give it to you.

Engagement models

Three shapes, and they are not interchangeable.

  • Project engagement. Bounded scope, defined deliverable, fixed or capped fee. The right structure for a single matter, a readiness assessment, a protocol negotiation, or a remediation. The defined deliverable is what keeps it from drifting into an open tab.
  • Advisory retainer. A monthly or annual commitment for availability, including the call you make when a hold question lands at 4pm on a Friday. Suited to organizations with recurring litigation and no in-house eDiscovery function. It is worth the cost only if you use it, so agree in advance what triggers a call.
  • Expert engagement. Retained for testimony, either consulting-only within the work-product protections or as a disclosed testifying expert. This is a different posture from the other two. Independence and documentation discipline matter more than responsiveness, and the engagement should be structured that way from the first day rather than converted out of an advisory relationship after a problem surfaces.

Cutting across all three is a question worth settling before the first call: who signs the engagement letter. A consultant retained directly by the company is doing business advice. A consultant retained by outside counsel, to assist counsel in rendering legal advice, sits inside a materially stronger claim to work-product protection, and a consulting expert not expected to testify is generally shielded from discovery absent exceptional circumstances under Rule 26(b)(4)(D). That structure is easy to put in place at the start and awkward to retrofit once the analysis exists in a folder somewhere. If any part of the engagement involves assessing what went wrong, route it through counsel.

Key takeaway

Buy the role you actually need. Advice, execution, software, and testimony are four separate purchases, and the most common expensive mistake is assuming that buying one of them gets you another.

The test that matters when choosing an eDiscovery consultant

A good eDiscovery consultant makes themselves less necessary over time. The first matter is expensive because every decision is being made for the first time. The second should cost less, because those decisions are now written down. By the third, your own team should be running the process and calling the consultant only for the genuinely novel questions.

If your third matter costs what the first one did, and you are having the same arguments about scope and custodians that you had two years ago, you did not buy consulting. You bought labor. The difference between the two shows up in one place: whether anything was left behind that your organization now owns.