There is a gap between what law school teaches and what a first-year litigator is actually paid to do, and discovery sits squarely inside it. You will graduate able to brief a case, write a memo, and recite the standard for summary judgment. Then you will arrive at a firm, be handed a set of interrogatories due in three weeks, and realise that nobody has ever shown you how to answer one.
This is not a personal failing and it is not rare. Discovery is where the majority of a junior litigator's billable hours go for the first two or three years, and it is the one core skill the profession has quietly decided to teach by osmosis. At large firms the osmosis sometimes works, because there are five layers of associates above you. At small and mid-size firms there is often nobody between you and a partner whose interest is in getting the assignment out the door.
What follows is a method for teaching it to yourself: the order to learn things in, the analytical frame that converts copied language into actual strategy, what each discovery device is genuinely for, and a ninety day plan you can run alongside a full caseload.
Why copying templates feels hollow, and why that instinct is correct
Almost every junior starts the same way. You open the firm's document management system, find the last set of responses in a similar case, and adapt. It produces a document. It does not produce understanding, and the discomfort you feel about that is diagnostic rather than neurotic.
There are two distinct problems with learning from precedent alone.
The first is that you absorb the shape without the reasoning. You learn that responses open with a block of general objections, that requests use the phrase "possession, custody, and control", that everything is served "subject to and without waiving the foregoing objections". You cannot say why. When opposing counsel challenges a response, or a partner asks you to justify one, you have nothing underneath the language.
The second problem is worse, because it is invisible. Firm templates accrete. The set you are copying was adapted from a set that was adapted from a set, and somewhere in that chain the rules changed. The December 2015 amendments to the Federal Rules of Civil Procedure are the clearest example. They rewrote the scope of discovery in Rule 26(b)(1) around proportionality, and they deleted the sentence every older template was built on, the one permitting discovery "reasonably calculated to lead to the discovery of admissible evidence". They also amended Rule 34 so that an objection must state its grounds with specificity, must state whether any responsive material is actually being withheld on the basis of that objection, and must specify a time for production.
A template written in 2012 and lightly edited ever since will teach you a standard that no longer exists and a response format that a well-prepared opponent can attack. Precedent is a valuable second step. It is a dangerous first one.
The reading order that actually works
Learn the authority before you learn the artefacts. Roughly a weekend of focused reading will put you ahead of a meaningful share of practitioners, including some senior ones.
Step one: read the rules, in full, in one sitting
In federal practice that means Rules 26 through 37, plus Rule 45 for non-party subpoenas and Rule 16 for the scheduling order that governs your deadlines. Read them straight through rather than looking things up as they arise. Discovery is a system, and the rules only make sense as one: Rule 26 sets scope and obligations, Rules 33 through 36 are the devices, Rule 37 is the enforcement mechanism that makes the rest of it real.
Then read the equivalent provisions in the state code where your case sits. State discovery regimes are not simplified copies of the federal rules. They diverge on numerical limits, response windows, whether leave of court is required, and, importantly, on required statutory formulations. California's Civil Discovery Act, to take the most heavily codified example, prescribes specific language for particular situations, including what a party must affirm when it cannot comply with a demand for production. Using the federal habit there is a losing move.
Step two: read the local rules, the standing orders, and the case-specific orders
This is the step juniors skip and judges notice. Your district or county will have local rules that cap discovery, prescribe the format of meet and confer letters, or require a pre-motion conference before anyone files a motion to compel. Individual judges publish standing orders that override the defaults in their courtroom. Your own case may have a protective order, an ESI protocol, and a Rule 502(d) order governing what happens when privileged material is produced by mistake.
Four documents therefore govern any given request: the federal or state rule, the local rule, the judge's standing order, and the case-specific orders. A request that complies with the first and violates the fourth is still wrong.
Step three: read the advisory committee notes
This is the highest-yield thing on this list and almost nobody does it. The Federal Rules are published with committee notes explaining what each amendment was meant to fix. They are available at no cost from the federal judiciary and from the major open-access law libraries. The notes to the 2015 amendments explain precisely why proportionality was moved into the definition of scope and why boilerplate objection practice was targeted. The notes to the 2020 amendment to Rule 30(b)(6) explain the new obligation to confer in good faith about the matters for examination.
Rules tell you what the standard is. The notes tell you what the drafters were worried about, which is what you will actually argue in front of a judge.
Step four: now read the templates, with the rules open
Having done the first three steps, you are no longer absorbing your firm's precedent. You are auditing it. That is a different and far more useful activity, and it will surface things worth asking about.
The element grid: how discovery stops being guesswork
Here is the single technique that converts "copying language" into strategy, and it takes about an hour per case.
Open the complaint and the answer. List every cause of action and every affirmative defence. For each one, write out its legal elements. Then build a grid: one row per element, with columns for the fact that proves it, who holds that fact, what document would contain it, and which discovery device retrieves it.
Every request you draft should trace to a cell in that grid. If a request does not trace to a cell, one of two things is true: it is a fishing expedition and should be cut, or you have found an element you have no plan to prove, which is a much more urgent discovery than anything in the document.
Take a trade secret claim under the federal Defend Trade Secrets Act, since it maps neatly onto the commercial and intellectual property work that fills junior dockets.
| Element | What must be proved | Where the proof lives | Device |
|---|---|---|---|
| Information qualifies as a trade secret | Independent economic value from not being generally known | Internal valuations, pricing analyses, competitor benchmarking | Requests for production; contention interrogatory late in the case |
| Reasonable measures to maintain secrecy | Confidentiality agreements, access controls, policies actually enforced | NDAs, onboarding records, access logs, IT policy documents | Requests for production; interrogatory to identify each measure taken |
| Not generally known or readily ascertainable | Absence of public disclosure | Disclosure history, third-party agreements, published materials | Interrogatory identifying every third party to whom the information was disclosed and on what terms |
| Misappropriation | Acquisition by improper means, or disclosure or use | Device forensics, transfer and download logs, exit interview records, messages | Requests for production of ESI; deposition of the departing employee; non-party subpoena to the new employer |
| Use in the defendant's business | The accused product or process incorporates the information | Design documents, source code, product roadmaps, launch timelines | Requests for production; interrogatory to identify each product incorporating the information |
| Damages | Loss to the plaintiff or gain to the defendant | Revenue records, customer lists, margin analyses | Requests for production of financials; interrogatory on the computation |
Once the grid exists, drafting is mechanical and, more importantly, defensible. When a partner asks why you propounded Request 14, the answer is not "it was in the template". It is "reasonable secrecy measures are a contested element, and their access logs are the only evidence of whether the controls they describe in the answer were actually enforced".
Print the grid. Physically tick elements off as the responses arrive. At the close of discovery, any unticked row is a hole in your case, and you will know about it while there is still time to fix it rather than at the summary judgment hearing.
What each device is actually for
Juniors tend to treat the devices interchangeably, asking every question through whichever form is in front of them. They have genuinely different purposes, and using the wrong one wastes a limited resource.
Interrogatories are for identification, not for facts
Interrogatories are answered by opposing counsel, not by the witness, and are therefore drafted to concede as little as language permits. You will never win a fact on an interrogatory. What you will get is a roadmap: the identity of witnesses, the custodians of documents, the systems where data sits, the dates of key events, the computation behind a damages number, and eventually the contentions supporting each claim.
They are also scarce. Federal practice caps them at twenty-five, including discrete subparts, and many states are stricter. Two disciplines follow. First, spend them on identification and save contention interrogatories, which ask a party to state the facts supporting a specific allegation, for later in the case when there are facts to be pinned to. Second, do not burn more than roughly two-thirds of your allotment on the first set. New issues surface, and an interrogatory you cannot serve because you exhausted the count on boilerplate is a self-inflicted wound.
One interrogatory to never omit is the expert identification interrogatory. Combined with the continuing duty to supplement, it is what prevents an expert appearing late in the case that you have had no chance to prepare for.
Requests for production are where the case actually lives
Documents and electronically stored information are what witnesses are examined about, what gets attached to dispositive motions, and what goes to the jury. This is the workhorse device and it deserves the most drafting attention.
The federal rule reaches material in the responding party's "possession, custody, or control", and the conventional request tracks all three words. It is worth understanding why, because it is a fair question whether the middle word does any work. The practical answer is that the three terms together foreclose definitional argument. The classic scenario is a party who does not physically hold a document and does not own it, but has the legal right to obtain it, medical records held by a treating provider being the standard example. Whether you call that custody or control matters less than that no reasonable reading leaves it outside the request. Some jurisdictions also prescribe the formulation by statute, in which case the debate is over before it starts: use the words the code uses.
Requests for admission are for elimination, not discovery
Admissions are the most misunderstood device. They are not a way to learn things. They are a way to remove things: to authenticate documents so you do not need a foundation witness, to establish undisputed background facts, and to narrow what has to be proved at trial.
They also carry the sharpest trap in the rules. In federal practice a request for admission is deemed admitted if it is not answered within the response period. That cuts both ways, and the direction it cuts depends entirely on your calendar discipline.
Depositions are for locking testimony
Written discovery tells you what the file contains. A deposition tells you what a human being will say under oath, and fixes it so it cannot move at trial. Federal practice presumptively limits each side to ten depositions and each deposition to one day of seven hours.
The organisational deposition deserves separate study. When you notice an entity rather than an individual, you specify the matters for examination and the organisation must designate someone to testify on its behalf, with that testimony binding the entity. Since the 2020 amendment, the parties must confer in good faith about those matters. Drafting the topics is a genuine skill, and it is worth reading a few well-drafted notices before you write your first.
Subpoenas and examinations
Non-party material comes by subpoena, which has its own service, notice, and geographic requirements, and its own protections for the recipient. Physical and mental examinations require a motion and a showing of good cause, and the condition must genuinely be in controversy. Neither is exotic, and both are easy to get procedurally wrong the first time.
The four jobs discovery is doing at once
Behind the mechanics, a competent litigator is pursuing four objectives simultaneously. Holding them in mind is what makes the difference between drafting a document and running a strategy.
- Learn. Find the facts you do not have, including the ones that hurt you. Discovering your own bad document in month four is a manageable problem. Discovering it at trial is not.
- Lock. Fix testimony and positions in place so they cannot be revised once the other side sees where the case is going. This is the primary function of depositions and contention interrogatories.
- Build the admissible record. Facts you know are worthless if you cannot get them in. Authentication, foundation, and business records go into the discovery plan from the beginning, not into a scramble before trial.
- Signal. Discovery is the first sustained demonstration of how well you know the file. Requests that quote the pleadings, name the right custodians, and reference documents by Bates number tell the other side that the case will be tried by someone who has read everything. That changes how they price settlement. It is not posturing, it is information.
Draft every request as though a judge will read it
Because one might. Every request you write is a potential exhibit to a motion to compel, read by a judge with a heavy docket who wants to resolve it quickly. Write requests that a judge can grant in a sentence.
- Tie the request to the pleading. A request that tracks the language of a specific paragraph of the complaint is very difficult to characterise as irrelevant.
- Bound it. Give a date range, name the custodians, identify the systems. Unbounded requests invite proportionality objections that judges increasingly grant.
- Define terms once and use them consistently. A definitions section is not decoration. It is what stops a responding party answering a narrower question than the one you asked.
- Be specific rather than exhaustive. This is counterintuitive and it is correct. A shotgun request makes an obstructive objection look reasonable. A narrow, document-specific request makes the same objection look like stonewalling. Serve a focused first set, read what comes back, and follow up with a targeted second set aimed at the gaps.
When you receive requests, run the mirror image of that analysis. Answer the question that was asked and not one adjective more, because every volunteered detail is a new line of examination at the deposition. Check whether your jurisdiction has construed the operative phrases, since terms like "describe in detail" and "identify" often carry a settled local meaning.
On objections, adopt a simple rule: never make one you would not be willing to stand up and argue. State the specific ground, explain briefly why it applies, and say plainly whether anything is being withheld because of it. General objections recited at the top of every response and incorporated by reference into every answer are, in a growing number of courts, treated as preserving nothing at all, and they signal that nobody thought about the requests individually. The reflex to object to everything and then produce everything "subject to and without waiving" is understandable when you are new and afraid of waiving something. It is also the habit most worth outgrowing, because a page of unexplained objections is not protection, it is an invitation.
Calendar discipline is the thing that actually ends careers
Substantive mistakes in discovery are usually recoverable. Missed deadlines often are not.
Untimely objections are waived unless the court excuses the failure for good cause. Requests for admission are deemed admitted if not answered in time, which can concede an element of a claim without anyone arguing it. And the obligation does not end at service: the duty to supplement is continuing, so an answer that was accurate in March can become a sanctionable answer in September if you learn something and say nothing.
The system that prevents all of this is unglamorous. Diary the response date the day the request arrives. Diary an internal draft date two weeks earlier. Get every extension in writing, from opposing counsel, before the deadline rather than after it. Nobody will teach you this either, and it is worth more than any drafting tip on this page.
A ninety day plan for an associate with no mentor
This runs alongside a full caseload. It is designed for someone who has roughly three to five hours a week to invest in their own competence.
Weeks one and two: the authority
Read the discovery rules in full, then the state equivalents, then your local rules and your judge's standing orders. Read the advisory committee notes to the discovery rules. Produce one page per device for yourself: the limits, the response window, the objection requirements, and the enforcement provision. That page is worth more than any purchased outline, because you wrote it.
Weeks three and four: what your court actually does
Pull ten motions to compel decided in your court, ideally by your judge, with the exhibits attached. Federal filings are broadly accessible through the public access system and through open dockets projects that mirror it at no cost. Read the request that was fought over, the objection, and the ruling. You will learn more about what your judge will grant in an afternoon of this than in a year of guessing, and you will see exactly which drafting choices cause requests to fail.
Month two: the grid and the why annotation
Build the element grid on one of your own live cases. Then adopt a rule for everything you draft: write a one-line justification next to every request and every objection before it goes out. If you cannot write the line, delete the item or go and learn it. Delete the annotations before service. Nobody else needs to see them, and the discipline is the entire point.
This is also the month to have the conversation that most juniors never initiate. Take a set you have drafted to whoever is supervising you and ask a narrow, specific question with your own answer already in it: "I objected to Interrogatory 7 on proportionality because it covers four years of a two year relationship. Would you have objected on burden instead, or is that the right ground?" A precise thirty second question with a proposed answer attached gets answered. "Can you teach me discovery" gets deflected, not out of malice but because it has no bounded response.
Month three: the live experience
Attend a deposition, even if you are only carrying documents. Take a short one yourself if a low-stakes witness comes up. Sit in on a discovery hearing, including someone else's. Watch a meet and confer, and notice how much of a dispute is resolved by tone rather than authority.
Ongoing: build your own precedent bank
Keep your own file of requests, objections, and responses that worked, each annotated with why it worked and what rule supports it. Within two years this will be more valuable than the firm's document management system, because the firm's system stores text and yours stores reasoning.
What to read when you want depth
A short, honest list. Most of it is available through a research subscription your firm already pays for.
- The rules and the advisory committee notes. Available at no cost from the federal judiciary and open-access law libraries. Start here before spending money on anything.
- Your jurisdiction's practice guide. Every developed jurisdiction has one work that practitioners actually cite. In California civil practice that role is filled by the Rutter Group practice guide on civil procedure before trial, which is available on Westlaw and covers both mechanics and strategy. In federal practice, Wright and Miller's Federal Practice and Procedure and Moore's Federal Practice serve the same function. Find the equivalent for your forum and read the discovery chapters properly.
- The Sedona Conference publications on electronic discovery. The Sedona Principles and the commentary on proportionality are the works courts themselves cite when deciding ESI disputes. If your practice involves any volume of electronic evidence, which now means all litigation, this is the most useful non-rule reading available.
- Federal Judicial Center materials. The pocket guides written for judges are unusually clear, and reading the guidance your judge received is an obvious advantage.
- Bar association continuing education. State bar young lawyer divisions and litigation sections run discovery programmes regularly, and the materials often include annotated sample documents. Ask your firm to cover the registration; the request is easy to justify.
- Decided motions to compel in your own court. The most jurisdiction-specific resource in existence and it costs nothing but time.
A note on research assistants and generative tools. They are genuinely useful for brainstorming the universe of requests an element might require, and for pressure-testing whether you have missed a category. They are not a source of authority. Anything a tool tells you about what a rule requires, what a phrase means, or what a court has held must be verified against the rule and the case before it goes into a document you sign. This applies with particular force to confident-sounding statements about drafting conventions, which are frequently drawn from marketing content rather than from the rules.
The mentorship problem, described accurately
It is worth naming the structural issue rather than reading it as a judgment on you. Most firms bill time and do not bill training. A partner who spends ninety minutes walking you through the reasoning behind a set of responses has spent ninety minutes that nobody pays for, on a task that would have taken them twenty minutes to do alone. Small firms feel this most sharply because there is no leverage to absorb it. The result is not indifference so much as an economic model with no line item for your development.
Three practical adjustments follow.
Stop looking to the partner for the teaching. Find the associate two years ahead of you, who still remembers being lost and knows the mechanics, and the associate six or seven years ahead, who knows the strategy. Those two people, asked specific questions, will teach you more than the partner will, and the ask costs them far less.
Use the paralegals. An experienced litigation paralegal often knows the local filing practice, the clerk's preferences, and the mechanics of production better than most lawyers in the building. Treat them as a primary resource rather than as support staff and they will save you repeatedly.
Calibrate your expectations honestly, because the most common reason capable juniors conclude they are failing is that they are measuring against the wrong timeline. At six months you should understand the mechanics and never miss a deadline. At one year you should be able to draft a full set from the pleadings without opening a template. At two to three years you start making the judgment calls that define real competence, which are almost always about what not to ask for. Nobody is fluent at eight months. The people who look fluent are three years in and had someone to ask.
One ethical line worth drawing early
Discovery is the most abused phase of civil litigation. Requests get served to impose cost, objections get made to impose delay, and both get billed. You will encounter this, and at some point someone will imply that a particular opponent does not merit real effort because they lack the resources to litigate a motion to compel.
Two things are true about that. It is an accurate description of how some practices operate. It is also not a defence available to you. In federal practice every request, response, and objection carries your signature, and that signature is a personal certification that the document is consistent with the rules, is not interposed for an improper purpose, and is neither unreasonable nor unduly burdensome. The sanction provision runs against the signer. There is no exception for a weak opponent, and the certification does not become less binding because the other side is unlikely to test it.
Set the line now, while you have no professional reputation to protect. It is much harder to move later.
How this looks outside the United States
Everything above describes American discovery, which is the most expansive party-driven disclosure system in the world. Lawyers trained elsewhere, or working on cross-border matters, should be clear that it is the exception rather than the norm.
In England and Wales the process is called disclosure and it is court-controlled. Outside the Business and Property Courts it runs under Part 31 of the Civil Procedure Rules; inside them it runs under Practice Direction 57AD, which took effect on 1 October 2022 and replaced the earlier disclosure pilot. That regime requires Initial Disclosure with the statements of case, then a structured negotiation over Extended Disclosure using defined models ranging from no disclosure at all to the broad search order, recorded in a Disclosure Review Document. There is no equivalent of serving twenty-five interrogatories as of right.
In India, discovery and inspection sit under Order XI of the Code of Civil Procedure, interrogatories generally require the leave of the court, and commercial disputes governed by the Commercial Courts Act framework carry a mandatory obligation to disclose documents on affidavit with real consequences for omission. Canada and Australia use documentary discovery combined with oral examinations for discovery, broader than the English model and considerably narrower than the American one.
Civil law jurisdictions largely do not have discovery in this sense. Document production is narrow, specific, and court-supervised, and the expectation that a party will hand over harmful material to its opponent does not exist. International arbitration has developed its own middle path, and the IBA Rules on the Taking of Evidence in International Arbitration, with the Redfern Schedule for resolving document requests, are the reference point most tribunals work from.
If your career is likely to cross borders, learn one system properly first. The analytical habit, tracing every request to an element you must prove, transfers everywhere. Only the procedural furniture changes.
Common questions from new litigators
Is it normal to be given discovery with no training?
It is the majority experience, and it is more pronounced at small and mid-size firms where there is no associate layer to absorb the teaching. The absence of training is a fact about firm economics, not a signal about your ability.
How long before discovery stops feeling opaque?
Roughly six months to be comfortable with the mechanics, about a year to draft confidently from the pleadings without a template, and two to three years before the strategic judgments feel natural. Associates who follow a deliberate method rather than absorbing by repetition move through that curve noticeably faster.
Should I read the rules or the templates first?
The rules, without exception. Templates read after the rules are a useful shortcut. Templates read before the rules teach superseded standards with no way to detect the error.
What is the single highest-value habit?
Writing a one-line justification next to every request and every objection before it leaves your desk. If you cannot write the line, you do not understand the item, and you have found exactly what to go and learn.
The short version
Discovery is learnable without a mentor. It requires reading the rules before the templates, understanding why each device exists rather than treating them as interchangeable forms, tracing every request to an element you have to prove, drafting as though a judge will read it, and building the calendar discipline that keeps procedural errors from becoming substantive losses.
The associates who become genuinely good at this are rarely the ones who received the best training. They are the ones who decided, early and deliberately, to understand why each sentence was there before they signed their name under it.
Related reading on LegalAlphabet
- Litigation paralegal careers in the US, the role that runs discovery logistics alongside you
- E-discovery analyst careers in the US, for the technical side of electronic evidence
- Document review attorney careers in the US
- Litigation support specialist careers in the US
- Entry level legal jobs for US law graduates
- Browse current legal jobs and litigation roles on LegalAlphabet
This article is professional development guidance for lawyers and does not constitute legal advice. Discovery rules vary by jurisdiction and change over time. Always verify the current text of the applicable rules, local rules, and standing orders before relying on any procedure described here.