What should you ask an eDiscovery specialist in an interview? Questions and what good answers sound like

An employer-side question bank for interviewing an eDiscovery specialist at a law firm: questions on collections, processing and productions with answer guides, a practical exercise, and the questions to keep out of the room.

The questions to ask when hiring an eDiscovery specialist are the ones that surface a pipeline a candidate has actually run — legal hold, collections, processing, hosted review, productions — plus how they guard client data, hold a production date, and know where the technical work ends and the lawyers' judgment begins.

Below is that question bank for law firms: what each question reveals, what strong answers sound like, a practical exercise, and the questions to keep off your list.

Before the interview

Write the questions from the job description you posted, not from a generic list.

If you have not settled the seat itself — which part of the discovery data pipeline the hire owns, what they hand to vendors, who they support day to day — start with our guide to how to hire an eDiscovery specialist, then sharpen the questions once your posting says what the hire will actually do.

The interview exists to test what that posting claims.

This page treats the seat as the data pipeline behind your discovery: preserving and collecting from custodians and business systems, processing into review platforms, supporting the review, producing to the agreed format — and being able to show, months later, where the numbers came from.

Decide which stretch of that pipeline you are hiring for and weight the questions toward it; a strong processing-and-productions hire and a strong collections-and-vendor hire are not the same experience.

Build the practical exercise from mock pipeline artifacts — a collection log with a gap in it, a processing exception report, two culling approaches for one data set — never from a live matter.

Decide the exercise and the scoring sheet before the first candidate sits down, keep both identical for every candidate, and score while the interview is fresh — a note written an hour after each interview beats a memory three weeks and six candidates later.

Two legal rails belong under the exercise.

Under EEOC guidance, a required test must be necessary and related to the job and must not exclude people of a particular protected group or people with disabilities.

And an employer must provide an accommodation a disabled applicant needs to apply for the job — a sign language interpreter, for example — unless it causes significant difficulty or expense.

Keep the exercise hypothetical.

The FLSA is the reason: "employ" includes "to suffer or permit to work" (29 U.S.C.

203(g)), so a candidate producing real, usable work for the firm — processing a live matter's data, for example — can be an employee owed wages.

A short exercise on a mock, disguised data set is the safer design.

Tell every candidate up front that they should not share client-identifying detail in the room: ask about the kind of matters and the data volumes, not the parties or the productions.

How a candidate handles the instruction is itself a data point — it previews exactly how they will talk about your clients.

And set the pay range before the loop starts — if you are hiring in California, Labor Code 432.3 requires giving an applicant the position's pay scale on reasonable request, whatever the firm's size.

Which skills questions reveal real eDiscovery specialist experience?

Take the last collection you personally ran. How did the custodian list and the data sources get decided, and how did the data actually move from each source into your processing environment?

What to listen for: Strong answers show scope being set with counsel rather than around them — who was in, who was out, and why — and a journey they can narrate step by step: what was pulled from mailboxes, phones, shared drives and business systems, and how each piece reached processing with its record intact.

Weak answers begin at the point where the data was already loaded.

Probe one source — how a phone collection actually ran is a good place to find the real detail.

Walk me through the last load you prepared for attorney review: what arrived, what you did to it before anyone could search it, and what never made it into the review set.

What to listen for: Strong candidates narrate decisions, not just steps: how duplicates were handled and on whose instruction, what happened to encrypted, corrupt or unsupported files, and how the exceptions were recorded so the record shows where every document went.

Weak candidates describe pushing buttons in a tool.

Ask what they showed the attorneys when a culling choice cut the set hard — the strong answer includes a conversation, not just a smaller number.

An attorney runs their terms and the hit report comes back far bigger than they expected. Take me through how you would help them read it.

What to listen for: Strong candidates sit next to the attorney and make the report legible: what the numbers actually count, where the terms overlap, how much of the volume is duplicate material, and what decision the report is there to inform.

Weak candidates reship the report and wait for a reaction.

The skill you are testing is translation — data explained to a lawyer, not a lawyer handed a spreadsheet.

What has to be true before you let a production leave the building — and if the other side challenged your counts six months later, what would you hand the partner?

What to listen for: Strong candidates answer in checks and records: the deliverable tested against the receiving party's specification, load files and metadata fields verified, exceptions resolved or documented — and a file they could pull months later that shows where each number came from.

The second half of the question separates a specialist from an operator: weak answers stop at I made the deadline, with nothing to hand anyone afterwards.

Where does your work start on a new matter — what do you do on the preservation and legal-hold side, and where does that hand off to counsel?

What to listen for: Strong answers describe coordinating with counsel rather than freelancing: what they preserve, what counsel decides about custodians and holds, and how they keep their own actions from altering anything while the matter is live.

Weak answers start at processing, as if preservation were someone else's problem until data lands.

Press on a time a hold touched a system they administer — the boundary they draw is the judgment you are hiring.

Which behavioral questions test judgment, confidentiality and deadlines?

Tell me about a matter where confidentiality rules shaped how you built the review workspace — separate matters, privilege screens, restricted users. How did you keep the walls straight?

What to listen for: Strong answers treat access as a design decision made before the first upload: who could see which matters, what was segregated and why, and how a new user was vetted mid-matter.

Weak answers add whoever asks to the platform and think about walls only after something leaks.

The seat holds several clients' case data at once, so score the walls above the walkthrough.

Tell me about a collection or processing job that came back far bigger than anyone planned — with the production date fixed. What did you do first?

What to listen for: Strong answers surface the mismatch early and in writing, give counsel the options with their consequences, and show what they re-sequenced to protect the date.

Weak answers are either I worked nights or the numbers were what they were, with no decision anywhere in them.

Ask what they told the attorneys on day one — the escalation habit is the point of the question.

Describe a time you found a gap in your own paper trail — a collection step nobody logged, an exception nobody recorded. What did you do?

What to listen for: Strong answers close the gap and change the habit: reconstruct what they can, tell the attorneys rather than quietly patching, and add the check that would have caught it.

Weak answers insist their records are complete.

A seat like this one lives on its paper trail — a candidate who can describe fixing their own documentation gap is more credible than one who claims to have none.

Tell me about a culling or de-duplication decision that had real consequences — something that got cut and later turned out to matter. How did you handle it?

What to listen for: Strong answers put the decision where it belongs — taken with counsel, on a stated approach, with the record showing what was applied — and describe what changed afterwards: the threshold revisited, the approach confirmed in writing, the sample re-checked.

Weak answers either own a decision they were never entitled to make alone or disown every number in the set.

You are listening for someone who implements judgment, not someone who improvises it.

Tell me about a time an attorney wanted to work in the review set before it was fully loaded or fully checked. What did you do?

What to listen for: Strong answers are serviceable without being unsafe: they gave the attorneys something to work in — a clearly marked partial set — while finishing the checks, or held the load and said why.

Weak answers are either nobody touches it until I am done, or everybody in whenever they ask.

Ask how they marked what was incomplete — strong candidates always know what state the data was in.

What practical exercise or test should an eDiscovery specialist complete?

Hand them a mock collection log for two or three custodians — dates, sources, file counts, one deliberate gap — and ask what they notice and what they would ask about before relying on it.

What to listen for: Strong candidates read the log the way they would have written it: what is missing, which entry does not reconcile with the others, and what they would confirm with the collector before defending any number in it.

Weak candidates take the log at face value and start planning from it.

The gap is the exercise — the habit you are testing is verifying the record before building on it.

Give them a mock processing exception report — encrypted files, corrupt files, unsupported types — and ask what they would do with each line and what the case team would hear about it.

What to listen for: Strong candidates triage rather than dump: what can be rerun, what needs a password or a decision from the client side, what goes to the attorneys in plain language, and how each line gets closed out in the record.

Weak candidates forward the report and move on.

Ask what they would do about the one corrupt custodian mailbox — that line hides a decision for counsel.

Give them two mock culling approaches for the same data set — de-duplicate and date-restrict versus load everything — and ask how they would lay out the tradeoffs for counsel.

What to listen for: Strong candidates frame it as counsel's decision and make the options legible: what each approach does to volume, what each risks cutting, and what the record would show either way.

Weak candidates argue for their own preference as if the choice were purely technical.

You are watching for a specialist who equips the decision instead of taking it.

Ask them to talk through setting up a review workspace for a new matter from a blank environment — structure, naming, and who gets access.

What to listen for: Strong candidates map access to the case team, decide naming and structure before documents arrive, and apply confidentiality rules from the first upload.

Weak candidates start with the software's features.

Ask who they would give access to mid-matter — a strong answer runs through the case team and counsel, not whoever asked.

Ask for the documentation behind one production they personally ran in the last year — the collection record, the processing exceptions, the pre-production checks — redacted before it reaches you, and have them walk you through how it would answer a challenge to the production's counts.

What to listen for: A strong candidate produces real documentation quickly and can trace any number in it back to a step they performed: how the material was collected, what was removed and on whose instruction, what was checked before it left.

A weak showing is an old summary someone else wrote, a final report with no backup behind it, or a production they cannot discuss in detail.

Redaction is part of the exercise — and so is being able to discuss the redacted document in detail.

What do strong vs weak answers sound like?

Collections: I scoped the custodians with counsel, kept a record of what was collected from whom and when, and I can still answer questions from it — strong. The vendor collected the data and I never saw the details — weak.

What to listen for: The strong version shows ownership of a record that outlives the matter — the candidate can go a level deeper on any line of it.

The weak version is not disqualifying because a vendor was involved; it is disqualifying because nothing was watched: no specification, no log, no visibility.

Collections you cannot see, you cannot defend.

Processing: duplicates and exceptions were handled the way counsel agreed, and I can show you where that was written down — strong. The software handles de-duplication, so I don't really touch it — weak.

What to listen for: The strong answer treats tool defaults as decisions somebody has to own — and names who owned them.

The weak answer outsources every consequential choice to the software's settings.

Ask what the default actually was and whether anyone chose it — a specialist can answer; an operator cannot.

Counts: every number in the production traces back to a report I still keep — strong. The counts came out of the platform; I never checked them — weak.

What to listen for: The strong answer is the defensibility habit the exercises test: records kept, reconciled, retrievable months later.

The weak answer is the one that turns a challenged production into a crisis — if nobody checked, nobody knows what the number is.

Probe by asking how long they keep the reports — the strong answer has a practice, not a shrug.

Judgment: I held the load until the privilege screen was applied and explained the delay to the case team — strong. I load whatever they send, however they send it — weak.

What to listen for: The strong answer shows the line between serviceable and safe: a clean load that arrives explained beats a fast one nobody can vouch for.

The weak answer sounds flexible and is the costly one — it is how privileged material ends up in a production nobody can explain.

Listen for whether the candidate can name the moment they decided to hold it.

Which questions are illegal or risky to ask?

Treat this as risk management, not a list of magic words.

Under EEOC guidance, federal law does not clearly forbid every interview question that touches race, sex, national origin, religion or age — but such questions can be used as evidence of an intent to discriminate unless they are justified by a business purpose, and the guidance's working rule is that pre-employment questions should be limited to what is essential to deciding whether a person is qualified.

An eDiscovery hire sits close to several of these questions — the whole interview is about whether this person can be trusted with sensitive material, and about their background — which is exactly why the script needs a clean list.

The flags below are what to keep out of it, and what to do instead.

Have employment counsel confirm the final version before your first interview.

  • Protected traits dressed as small talk. When did you graduate, are you planning children, what church do you attend, who watches your children — the EEOC publishes separate guidance on pre-employment inquiries about race, height and weight, financial information, unemployed status, religious affiliation, citizenship, and marital status or number of children. If a question would matter to you only because of a protected trait, it is not essential to the hiring decision.
  • Medical questions before an offer. FTC and EEOC guidance is blunt: employers should not ask medical questions before a conditional job offer. Ask about the job's demands — heavy production weeks, after-hours processing windows, trial support — and keep health conditions and disabilities out of the question script.
  • Salary history, asked directly. What were you making at your last firm? is barred or restricted in California, New York, Illinois, Massachusetts, Connecticut, Virginia and Nevada: California Labor Code 432.3 bars all employers, of any size, from seeking an applicant's salary history — compensation and benefits included — orally or in writing, personally or through an agent, and from relying on it in hiring and pay decisions. New York Labor Law 194-a bars employers from relying on an applicant's wage or salary history, from requesting or requiring it as a condition of being interviewed, considered, hired or promoted, or from seeking it from a current or former employer. Illinois (820 ILCS 112/10(b-5) and (b-10)) makes it unlawful to screen applicants by salary history, to request it as a condition of an interview or offer, or to seek it from a current or former employer. Massachusetts (G.L. c.149 s.105A(c)(2)) bars seeking it from the candidate or a current or former employer. Connecticut bars asking about wage and salary history unless the applicant volunteers it. Virginia (Virginia Code 40.1-28.7:12) and Nevada (NRS 613.133) also ban seeking salary history.
  • Salary history — what you can do instead. Ask expectations, not history: California Labor Code 432.3 expressly allows asking an applicant about their salary expectation for the position, and an employer may consider salary history the candidate volunteers without prompting. New York allows confirming history only at the offer stage — when you have made an offer with compensation and the candidate responds by citing prior pay to support a higher number. Massachusetts allows confirmation only after voluntary disclosure or after an offer with compensation has been made.
  • Criminal history, asked too early. It is generally not illegal for an employer to ask about an applicant's background or require a background check, apart from restrictions around certain medical and genetic information — but when and how you ask is regulated, and the FTC and EEOC guidance warns that a criminal-record exclusion that significantly disadvantages a protected group and does not predict job performance can be unlawful disparate impact. California's Fair Chance Act (Gov. Code 12952) bars employers with five or more employees from including conviction-history questions on an application, or otherwise inquiring into or considering conviction history, until after a conditional offer. New York City's Fair Chance Act, part of the NYC Human Rights Law since October 27, 2015, makes it unlawful for most employers to ask about or consider criminal history before a conditional offer. Illinois' Job Opportunities for Qualified Applicants Act (820 ILCS 75/15) bars covered employers from asking until the applicant has been found qualified and notified they are selected for an interview — or, when there is no interview, until after a conditional offer.
  • The step after a conviction question becomes lawful. In California, an employer that plans to deny a job over conviction history must make an individualized assessment, notify the applicant in writing, and give at least five business days to respond before a final decision.
  • A practical exercise built on the firm's real work. The FLSA defines employ as including to suffer or permit to work (29 U.S.C. 203(g)), so a candidate producing real, usable work for the firm can be an employee owed wages. An exercise run on a live matter also hands the candidate client data before they are hired. Keep the exercise on a mock, disguised data set.
  • Assuming size exempts you. Federal anti-discrimination law covers private employers with 15 or more employees for Title VII, the ADA and GINA, and 20 or more for the age discrimination law (ADEA) — in each case counting employees who worked at least 20 calendar weeks in the current or prior year — and the protections reach applicants and former employees as well as current staff. The federal numbers are not the whole test: California's Fair Chance Act reaches employers with five or more employees, and this page's research did not confirm a size threshold for every state and city statute named above. Confirm coverage with employment counsel.
  • Treating this page as the whole map. This page names only the jurisdictions its research covered; it is not a count of the states and cities with salary-history or fair-chance rules, and many cities and other states have their own. Before you finalize your script, have employment counsel check the rules where your firm and each candidate sit.

After the interview

Score the same day, on the same axes.

Pipeline depth, defensibility of the counts, discretion, deadline discipline — the four things these questions were built to test.

Grade each interview alone before the panel compares notes — a panel that talks first grades to the strongest voice, not the strongest evidence.

What strong answers sound like.

Records you could pull months later: custodians scoped with counsel, collections logged, duplicates and exceptions handled on a stated approach, productions reconciled before they left.

Strong candidates volunteer the stage that went sideways and what they changed — and in the exercises they say plainly what they have never touched, then reason their way to a next step.

An honest map of their own limits is worth more than a confident tour.

Verify before the offer.

Write down the specifics you intend to check — the employers, the platforms, the matters as the candidate described them — and call the references who actually watched the work, the supervising attorney first.

Keep what you wrote down: FTC and EEOC guidance says private employers must keep personnel and hiring records, including applications from people who were not hired, for at least one year from the date the record was made or the personnel action taken, whichever is later.

Know your market before the pay conversation.

Our research found no BLS occupation code for eDiscovery specialists — none of the BLS series we read measures this title by name — so the pay benchmarks here are the broader occupations that bracket it in the legal services industry: Legal Support Workers, All Other (SOC 23-2099), with a median annual wage of $63,100, and Computer Occupations, All Other (SOC 15-1299), with a median annual wage of $105,500 — both BLS OEWS May 2025.

Employers may report eDiscovery-type staff under paralegals (SOC 23-2011) as well, and our research cannot tell you which code a given employer actually uses, so read the two medians as brackets, not one number.

The questions an eDiscovery specialist candidate asks you — and what they signal.

Leave real room at the end — the questions are evidence too:

  • "Where does the pipeline break down today — collections, processing, or review support?" — a candidate who asks where the operation hurts is already thinking about their first 90 days. Answer honestly; the person you hire will inherit the bottleneck, and a good candidate is sizing it.
  • "Which platforms would I work in, and who owns the vendor relationships?" — separates running the environment from working inside one, the same distinction the skills questions above keep circling. Name your stack and the gaps you expect the hire to close.
  • "How is a matter staffed — do I own the pipeline end to end, or do paralegals and litigation support own parts of it?" — a boundary question, and a smart one: the seam between this seat and your legal staff is where discovery work goes sideways. Answer from your current structure, not the one you hope to build.
  • "What stays in-house, and what goes to providers?" — signals a candidate who has managed vendors and wants to know whether the job is running a pipeline or supervising one. Whatever your answer is, make it honest about the split.
  • "What is the pay range for this role?" — a fair question, and in some places one with a legal answer behind it: California requires employers to give an applicant the position's pay scale on reasonable request, and Nevada requires the wage or salary range for an applicant who has completed an interview. Have the number ready and consistent with the posting.
  • No questions at all — do not read it as disinterest on the spot; ask what they are still weighing and what would make the role an easy yes. The answer to that is data too.

This page is employer information, not legal advice. Interview-question rules come from EEOC guidance and from state and city statutes that differ and change; have employment counsel review your question script and exercise before you use them.

More hiring resources

Hiring an eDiscovery specialist?

Post the opening where eDiscovery specialists look for law firm jobs — the pipeline, the platforms and the pay range up front.