Checklist · Litigation

Moving for Class Certification: A Practical Checklist

Type
Checklist
Jurisdiction
National
Date
2024-04-12
Topic
Litigation

Class certification is the whole ballgame: certify and the case settles, lose and the case usually dies. This checklist walks the plaintiff-side motion from the first draft of the class definition through the Rule 23(f) petition. It covers defining a class that is precise, objective, ascertainable, and not fail-safe; screening out the uninjured-members problem that TransUnion put at the center of modern practice; vetting representatives and counsel under Rule 23(a) and 23(g); and choosing the right Rule 23(b) category. It then turns to the evidentiary core of the motion: proving predominance with common evidence, building a damages model that matches the liability theory under Comcast, surviving the Daubert gate, sequencing certification discovery, and writing a trial plan the judge can actually picture. Each phase ends with the traps that sink otherwise strong motions. Because Rule 23 doctrine is genuinely unsettled on ascertainability, uninjured members, and expert gatekeeping at certification, the circuit-specific warnings throughout are not boilerplate — check them before you file. The document closes with a consolidated one-glance checklist and a Primary Authority section.

Moving for Class Certification: A Practical Checklist

Litigation > Class Actions | Published 12 April 2024 - Updated 16 September 2025 | UpLaw - uplaw.us

Certification is not a procedural formality. It is the moment a case stops being one person's grievance and becomes a bet-the-company exposure — and the moment a judge decides whether that transformation is fair. Win it and the defendant's settlement calculus changes overnight. Lose it and you are left with a plaintiff whose claim is worth less than the filing fee.

Keywords: class certification, Rule 23, predominance, commonality, typicality, adequacy, ascertainability, fail-safe class, Comcast, TransUnion, uninjured class members, Rule 23(f), class counsel, Daubert, trial plan

How to use this checklist

This is the plaintiff's playbook for a contested Rule 23 motion in federal court. Work the phases in order — they are sequenced the way a well-run case is sequenced, not the way the rule is numbered. The class definition comes first because everything else is downstream of it; the damages model comes before the brief because a Comcast failure cannot be papered over with rhetoric.

For the doctrinal foundations behind every item here, read Class Action Litigation: A Comprehensive Guide to Rule 23. For the full map of the practice area — pleading, certification, settlement, appeal — see The Class Action Toolkit. And before you file, read your opponent's script: Opposing Class Certification: A Defense Checklist is the mirror image of this document, and every argument in it is one you should be pre-empting.

Running example. Throughout, we follow Whitcomb v. Verdant Row Naturals, Inc., a fictional consumer case in the Northern District of California. Verdant Row sells a supplement, MetaboClear, whose label says "clinically proven to burn fat." Named plaintiff Dana Whitcomb bought two bottles at a grocery store. Her counsel at the fictional firm Ardsley & Poe want to certify a California statewide class of purchasers. We will watch them break the class definition, fix it, and build the record.

Phase 1 — Define the class: precise, objective, ascertainable, and not fail-safe

  • [ ] Draft the class definition before the complaint is finalized, and expect to amend it at certification (Rule 23(c)(1)(B) requires the certification order to define the class, and Rule 23(c)(1)(C) lets the court alter or amend it before final judgment).
  • [ ] Use objective criteria only: who bought what product, in what geography, during what period, through what channel. Never use a criterion that requires a merits finding or a look inside someone's head.
  • [ ] Screen for a fail-safe class — a definition whose membership turns on liability ("all persons who were deceived by," "all employees who were misclassified," "all borrowers whose fees were unlawful"). Fail-safe classes are one-way ratchets: if plaintiff loses, no one is bound, so no one was ever in the class. Most courts to confront the problem either decline to certify such a class or redefine it under Rule 23(c)(1)(C), though the circuits differ on whether "fail-safe" is a freestanding bar or simply an aspect of the definiteness and predominance inquiries. Do not rely on a court's willingness to rewrite your definition for you.
  • [ ] Screen for a class that requires mini-trials to identify members ("all persons who saw the advertisement," "all customers who relied on the representation").
  • [ ] Confirm the named plaintiff is inside the class as defined. It is embarrassingly common for a redraft to define the representative out.
  • [ ] Fix the class period to the applicable limitations period, adjusted for any tolling (including American Pipe tolling if a prior class action was filed).
  • [ ] Identify subclasses now if different products, states, statutes, or time periods carry materially different elements. Rule 23(c)(5) permits subclasses, each treated as a class.
  • [ ] Know your circuit's ascertainability law. It is genuinely split: the Third Circuit requires a showing of "administrative feasibility" for identifying members (Carrera v. Bayer Corp., 727 F.3d 300 (3d Cir. 2013); Marcus v. BMW of North America, LLC, 687 F.3d 583 (3d Cir. 2012)), while the Seventh and Ninth Circuits have rejected any heightened requirement beyond a clear, objective definition (Mullins v. Direct Digital, LLC, 795 F.3d 654 (7th Cir. 2015); Briseno v. ConAgra Foods, Inc., 844 F.3d 1121 (9th Cir. 2017)). Where you file matters enormously in low-dollar consumer cases where nobody keeps receipts.

Why it matters, and the traps

The class definition is the load-bearing wall. Predominance, ascertainability, typicality, notice, preclusion, and the scope of any judgment all rest on it. A sloppy definition is not a drafting problem you clean up later — it is a substantive defect the defense will spend forty pages exploiting, and it is the single most common reason a first certification motion fails.

Worked example. Ardsley & Poe's complaint defined the class as "all California consumers who purchased MetaboClear and were deceived by Verdant Row's false 'clinically proven' claim." That is a textbook fail-safe class in two respects: membership turns on whether the claim was false (a merits question) and on whether the individual consumer was deceived (an individualized state-of-mind question). Verdant Row's opposition would have written itself.

The fix: "All persons who, between March 1, 2020 and the date of class certification, purchased MetaboClear in California for personal use and not for resale." Every criterion is objective and verifiable. Deception is now proved classwide through California's reasonable-consumer standard rather than smuggled into the definition. Resellers are excluded because their damages theory differs. The class period tracks the four-year UCL limitations period. Ardsley & Poe should also carve out the standard exclusions: Verdant Row's officers and directors, the presiding judge and chambers staff, and persons who timely opt out.

Phase 2 — Screen out the uninjured-members problem early

  • [ ] Ask the hard question: does every person inside your definition have a concrete Article III injury? Under TransUnion LLC v. Ramirez, 594 U.S. 413 (2021), "[e]very class member must have Article III standing in order to recover individual damages," and a bare statutory violation without concrete harm will not do.
  • [ ] Count the uninjured. If your model concedes that some appreciable share of the class bought at a discount, never paid the challenged fee, would have bought anyway, or suffered no loss, know the number and have an answer.
  • [ ] Decide which of the three standard answers you are using: (1) narrow the definition to exclude the uninjured; (2) show the uninjured share is de minimis and administratively separable at the claims stage; or (3) show injury itself is a common question provable classwide.
  • [ ] Do not let your own expert create the problem. A regression that shows zero overcharge for a subset of purchasers hands the defense its best brief.
  • [ ] Know your circuit. The D.C. Circuit in In re Rail Freight Fuel Surcharge Antitrust Litigation (D.C. Cir. 2013) and the First Circuit in In re Asacol Antitrust Litigation (1st Cir. 2018) treat a non-trivial population of uninjured members as fatal to predominance where winnowing them out would require individualized inquiry; other circuits are more forgiving. The Supreme Court took up the question in Laboratory Corporation of America Holdings v. Davis and then dismissed the writ as improvidently granted in June 2025, leaving the split in place. Verify the current state of the law in your circuit before you file.

Why it matters, and the traps

TransUnion moved the uninjured-members fight from a footnote to the center of modern certification practice. The defense will not argue that your theory is wrong; it will argue that your theory, even if right, sweeps in thousands of people who lost nothing — and that sorting them out is a job for thousands of individual hearings, which is the definition of a predominance failure.

The trap is complacency in statutory-damages cases. Counsel who assume that a statutory violation is per se injury are litigating a pre-2021 world. Plead and prove the downstream concrete harm — the disclosure that actually went to a third party, the money actually paid, the credit actually denied — not just the violation.

Worked example. Verdant Row's expert will argue that 18% of MetaboClear buyers were repeat purchasers who kept buying after using the product, which "proves" they got what they paid for. Ardsley & Poe's answer is not to fight over the 18%; it is to reframe: the injury is the price premium paid at the register for a label claim that was worthless, and a repeat purchaser paid that premium twice. Injury is a common question because the premium is embedded in the market price. That answer only works if the damages model in Phase 8 is built to support it.

Phase 3 — Select and vet the class representatives

  • [ ] Confirm the representative's claim arises from the same event, practice, or course of conduct and is based on the same legal theory as the class's — that is typicality under Rule 23(a)(3).
  • [ ] Run the unique-defense test. A representative subject to a defense that does not apply to the class (an arbitration agreement, an individualized release, a statute-of-limitations problem, actual knowledge of the truth) is atypical and will become the case's center of gravity.
  • [ ] Vet credibility. Adequacy under Rule 23(a)(4) is where courts park credibility concerns. A representative with a felony fraud conviction, a history of serial plaintiffing, or shifting deposition testimony is a target.
  • [ ] Confirm the representative understands the case and the fiduciary role — not the legal theory in detail, but the basic nature of the claim, the duty to absent members, and the fact that she cannot settle only her own claim.
  • [ ] Check for conflicts between the representative and any segment of the class. Amchem Products, Inc. v. Windsor, 521 U.S. 591 (1997) is the canonical warning: a single representative cannot fairly serve both the presently injured and the exposure-only future claimants.
  • [ ] Recruit more than one representative where the class spans products, time periods, or purchase channels. Redundancy is cheap insurance.
  • [ ] Prepare the representative for a deposition that is really an adequacy examination. See Preparing to Take a Deposition: A Practical Checklist for the mechanics from the other side of the table.

Why it matters, and the traps

Typicality and adequacy are low bars in theory and lethal in practice, because they are the only Rule 23 requirements that turn on facts entirely within the defense's power to develop. The defense cannot depose predominance. It can depose your client.

The most under-managed trap is the arbitration clause. If the named plaintiff clicked through terms containing a class-action waiver and the class members did not (or vice versa), you have a typicality problem, a standing-to-represent problem, and an individualized-issues problem all at once. Diligence the enrollment records before you name the plaintiff, not after. Our companion Drafting an Arbitration Clause: A Practical Checklist shows how sophisticated defendants build these waivers precisely to defeat aggregation.

Worked example. Ardsley & Poe named Dana Whitcomb, a grocery-store purchaser, but 40% of MetaboClear sales came through Verdant Row's website, where buyers accepted terms containing an arbitration clause with a class waiver. Whitcomb is not bound by it — she bought at retail. That is a strength, not a weakness: the firm should define the class to exclude direct web purchasers (or subclass them) rather than certify a class half of whose members are headed to individual arbitration. One clean class of 60% of the sales beats a doomed class of 100%.

Phase 4 — Vet class counsel under Rule 23(g)

  • [ ] Prepare the Rule 23(g)(1)(A) showing on all four statutory factors: (i) the work counsel did identifying and investigating the claims; (ii) counsel's experience in class actions, complex litigation, and the specific claim type; (iii) counsel's knowledge of the applicable law; and (iv) the resources counsel will commit.
  • [ ] Submit a firm-resume declaration with prior certification and trial results, not just settlements.
  • [ ] Disclose any fee agreement and any litigation-funding arrangement your district's local rules or standing orders require.
  • [ ] Address Rule 23(g)(4)'s command that class counsel "must fairly and adequately represent the interests of the class."
  • [ ] If more than one firm has filed, resolve leadership early — by agreement, or by competing applications — rather than letting the court do it under Rule 23(g)(2).
  • [ ] Sanity-check capacity. A judge who suspects your firm cannot fund a two-week trial with three experts will find that fact under 23(g) rather than under adequacy.

Why it matters, and the traps

Rule 23(g) is the least-briefed and most-underestimated certification requirement. It is also the one that gives a skeptical judge an easy, appeal-proof way to say no. Do not bury it in a footnote. Two well-supported pages with a declaration will do.

The trap is the appearance of a lawyer-driven case. If the representative was recruited by an advertisement, cannot describe the product she bought, and defers every question to counsel, the court will read Rule 23(a)(4) and 23(g) together and see a class action in search of a plaintiff.

Phase 5 — Build the Rule 23(a) record

  • [ ] Numerosity (23(a)(1)). Prove impracticability of joinder with actual evidence — sales records, employee rosters, transaction data — not a guess. There is no magic number, but classes of 40+ are routinely found numerous and classes under 20 routinely are not. Also address geographic dispersion, small individual claim value, and the identifiability of members.
  • [ ] Commonality (23(a)(2)). Frame the common question as a common contention capable of classwide resolution, whose determination "will resolve an issue that is central to the validity of each one of the claims in one stroke." That is Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), and the emphasis is on the common answer, not the common question.
  • [ ] Identify the glue: the uniform policy, the uniform label, the uniform contract, the uniform algorithm. Dukes failed because Wal-Mart's policy was one of discretion — millions of independent decisions with no common driver.
  • [ ] Typicality (23(a)(3)) and adequacy (23(a)(4)): carry forward the Phase 3 work with declarations, not argument.
  • [ ] Assemble the evidence into a rigorous-analysis record. Under General Telephone Co. of the Southwest v. Falcon, 457 U.S. 147 (1982) and Dukes, the court must conduct a rigorous analysis that will frequently overlap with the merits. Certification is not decided on the pleadings.
  • [ ] Understand the limit of merits inquiry: Amgen Inc. v. Connecticut Retirement Plans & Trust Funds, 568 U.S. 455 (2013) holds that a merits question may be left for trial when it is itself common to the class — materiality in a securities case is a common question, so it need not be proved at certification.

Why it matters, and the traps

The gap between "we all have the same question" and "we will all get the same answer from the same evidence" is where most commonality arguments die. Dukes did not raise the numerosity or typicality bar; it raised the proof bar. Bring the policy document, the training deck, the uniform script, the source code. Bring the corporate designee testimony admitting the practice was uniform.

The complementary trap is over-reading Dukes into an obligation to try the merits at certification. Amgen is your shield: you must prove the Rule 23 prerequisites by a preponderance, but you do not have to win the case to certify it.

Phase 6 — Choose the Rule 23(b) category

  • [ ] Rule 23(b)(1)(A) — inconsistent adjudications would establish incompatible standards of conduct for the defendant. Narrow; think ratemaking, trust administration, plan-wide fiduciary duties.
  • [ ] Rule 23(b)(1)(B) — limited fund. Also narrow, and tightly policed by Ortiz v. Fibreboard Corp., 527 U.S. 815 (1999), which requires a fund whose limit is established independently of the parties' agreement.
  • [ ] Rule 23(b)(2) — the defendant "has acted or refused to act on grounds that apply generally to the class," so that final injunctive or declaratory relief is appropriate respecting the class as a whole. Use it for indivisible relief: change the label, stop the practice, reform the plan.
  • [ ] Do not use (b)(2) as a vehicle for individualized money damages. Dukes held that claims for monetary relief may not be certified under (b)(2), "at least where (as here) the monetary relief is not incidental to the injunctive or declaratory relief." Backpay was not incidental. Neither is your restitution claim.
  • [ ] Rule 23(b)(3) — predominance and superiority. This is the damages workhorse, and it carries opt-out rights and Rule 23(c)(2)(B) individual notice as the constitutional price of binding absent members (Phillips Petroleum Co. v. Shutts, 472 U.S. 797 (1985)).
  • [ ] Consider hybrid certification: a (b)(2) class for the injunction plus a (b)(3) class for damages, or a (b)(3) class with a Rule 23(c)(4) issue class carving out the common liability issues for classwide trial and leaving damages for individual proceedings.
  • [ ] Budget for notice. Under Eisen v. Carlisle & Jacquelin, 417 U.S. 156 (1974), the plaintiff bears the cost of Rule 23(c)(2)(B) notice in a (b)(3) class. The 2018 amendments permit electronic means, which has made this dramatically cheaper — but it is still real money, and the court will ask who is paying.

Why it matters, and the traps

Category selection is strategy, not taxonomy. A (b)(2) class avoids notice costs and opt-outs; a (b)(3) class delivers the money. Plaintiffs who try to get both by dressing damages up as "incidental" relief are running straight into Dukes.

The underused tool is Rule 23(c)(4). If your predominance problem is entirely on the damages side, ask the court to certify liability issues — duty, breach, the existence of the uniform practice, the falsity of the representation — for a single classwide trial, and to reserve individualized damages for claims administration or follow-on proceedings. Circuits differ on how freely 23(c)(4) may be used to sidestep 23(b)(3) predominance, so pitch it as a fallback in the alternative, and support it with a trial plan (Phase 11).

Phase 7 — Predominance: name the common evidence for every element

  • [ ] Build an element-by-element matrix. Down the left, every element of every claim (and every affirmative defense the defense has actually pleaded). Across the top: Common evidence? Individualized evidence? Which exhibit or expert proves it?
  • [ ] For each element, name the specific proof: the uniform label, the standardized contract, the transaction database, the regression, the survey.
  • [ ] Confront reliance head-on. Consumer protection is state law, and it varies materially — this is one place where you cannot reason from a nationwide rule. Some states permit classwide proof through an objective standard (California's "reasonable consumer" test under the UCL and CLRA is the familiar example); others require proof of individualized reliance or causation, which is often fatal to predominance. Research the specific statute of every state in your class and say so in the brief. In securities cases, invoke the fraud-on-the-market presumption of Basic Inc. v. Levinson, 485 U.S. 224 (1988) — and be ready for the defendant's price-impact rebuttal, which Halliburton Co. v. Erica P. John Fund, Inc., 573 U.S. 258 (2014) permits at the certification stage, as refined by Goldman Sachs Group, Inc. v. Arkansas Teacher Retirement System, 594 U.S. 113 (2021). See Securities Fraud Litigation: A Comprehensive Guide to Rule 10b-5 Private Actions.
  • [ ] Confront variation in state law if you are certifying a multistate class. Group the states by element, submit a state-law variation appendix, and propose subclasses. A nationwide class under fifty consumer statutes is a manageability argument gift-wrapped for the defense.
  • [ ] Where individual proof is unavoidable for a discrete element, use representative or statistical evidence — but only where each class member could have relied on that same evidence to prove her own individual case. That is the rule of Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442 (2016), which approved a time study in a donning-and-doffing case because the employer's failure to keep records made representative proof the only proof available.
  • [ ] Distinguish Tyson Foods from Dukes's rejected "Trial by Formula": statistical evidence may fill an evidentiary gap the defendant created; it may not manufacture a common answer where the underlying conduct was individualized.

Why it matters, and the traps

Predominance is not a headcount of common versus individual issues. It is a qualitative judgment about whether the case can be tried on common proof. The matrix is the tool that turns that judgment into a document — and it doubles as the skeleton of your trial plan and your brief.

The most damaging trap is conceding an individualized element by silence. If the defense identifies an element you never addressed — causation, exposure, damages, an affirmative defense — the court will assume it requires individual proof. Address every element, including the ones you would rather not discuss.

Phase 8 — The damages model must match the liability theory

  • [ ] Retain the damages expert early, before the class definition is locked. The model and the definition must be built together.
  • [ ] Confirm the model measures only damages attributable to the theory certified. That is the holding of Comcast Corp. v. Behrend, 569 U.S. 27 (2013): the plaintiffs' regression measured the combined effect of four antitrust theories after the district court certified only one, and the model was therefore incapable of measuring classwide damages.
  • [ ] Pressure-test the model against every theory the court might certify. If the court certifies a narrower theory than you pleaded, does the model still work? If not, you have built a Comcast problem into your own case.
  • [ ] Confirm the model is capable of measurement across the entire class on a common methodology — even if the individual numbers must later be computed from data.
  • [ ] Know the limit: Comcast does not require that every class member's damages be identical or that damages be calculable to the dollar at certification. Individualized amounts generally do not defeat predominance; an unworkable methodology does.
  • [ ] Choose the right model for the theory: a price premium / conjoint analysis for a mislabeling claim; a full refund where the product was worthless or illegal to sell; a benefit-of-the-bargain measure for a contract claim; a regression overcharge in an antitrust case; an event study in securities. For the underlying doctrine, see The Litigation Damages Toolkit: Valuing and Proving Money Awards.
  • [ ] Have the expert actually run the model on real data — a working demonstration on a sample beats a promise that the model "can be applied."

Why it matters, and the traps

Comcast is the single most-cited weapon in the defense arsenal, and the mismatch it condemns is almost always self-inflicted. Plaintiffs plead five theories to survive a motion to dismiss, then build one damages model to cover all five, then watch the court certify two of them and reject the model as overinclusive.

Worked example. Ardsley & Poe's theory is that the "clinically proven" claim commanded a price premium. Their expert, Dr. Priya Ramachandran, runs a conjoint survey isolating consumers' willingness to pay for that specific attribute, then multiplies the resulting premium by the number of units sold in California — a figure available directly from Verdant Row's own sales database. The model measures the premium attributable to the challenged claim and nothing else. Critically, it does not measure a "the product doesn't work at all" full-refund theory, so Ardsley & Poe must either drop that theory or build a second model for it. Pleading both and modeling one is exactly the Comcast trap.

Phase 9 — Expert evidence and the Daubert gate at certification

  • [ ] Assume the court will apply Rule 702 and Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993) to certification experts. Most circuits require at least a full Rule 702 analysis of expert opinions that are essential to a Rule 23 prerequisite; some frame the inquiry more loosely. Build to the strictest standard.
  • [ ] Remember the December 2023 amendments to Rule 702, which clarify that the proponent must show each element by a preponderance and that the expert's opinion must reflect a reliable application of the methodology to the facts. Cite the current rule text, not pre-2023 formulations.
  • [ ] Have the expert produce a complete Rule 26(a)(2)(B) report, with data, code, and workpapers — the defense will ask, and Rule 37(c)(1) preclusion for withheld materials is a real risk. See Rule 37(c)(1) and the Preclusion of Undisclosed Evidence: A Practical Guide.
  • [ ] Read your own expert's report adversarially, or have a colleague do it. Then read Challenging an Expert Under Daubert and Rule 702: A Practical Checklist and run every attack in it against your own witness before the defense does.
  • [ ] Prepare for the battle of the experts problem. Courts may not simply pick a side on the merits at certification, but they must resolve expert disputes that bear on whether Rule 23 is satisfied. Frame your expert's opinion as answering the Rule 23 question (is damages measurement possible on a common methodology?), not the merits question (what is the number?).
  • [ ] Consider a rebuttal expert to defuse the defense's uninjured-members and model-mismatch attacks rather than leaving them to argument.

Why it matters, and the traps

An excluded expert at certification is usually a denied motion. There is no reserve tank: if your damages model is stricken, you have no common methodology, and predominance fails.

The recurring trap is the expert who over-claims. An economist who says "my model proves every class member was injured" invites a Daubert fight she does not need to have. The right opinion is narrower and far more durable: "the price premium attributable to the challenged claim can be reliably estimated on a classwide basis using defendant's own transaction data and standard conjoint methodology."

Phase 10 — Certification discovery

  • [ ] At the Rule 26(f) conference, propose an explicit certification-discovery schedule with a briefing sequence and expert deadlines. Do not let the court bifurcate discovery in a way that starves you of merits facts you need for the rigorous analysis. See Rule 26 Initial Disclosures and the Discovery Plan: A Practical Checklist.
  • [ ] Resist a hard merits/certification bifurcation. Because Dukes requires a rigorous analysis that overlaps the merits, certification discovery necessarily reaches merits facts. Say so, in writing, in the discovery plan.
  • [ ] Get the transaction data: the full sales database, in native format, with a data dictionary and a Rule 30(b)(6) witness who can authenticate it. Numerosity, ascertainability, and the damages model all run on it.
  • [ ] Get the uniformity evidence: the policy manuals, the label approvals, the marketing briefs, the A/B tests, the internal complaints, the substantiation file.
  • [ ] Take a Rule 30(b)(6) deposition on the uniformity of the practice and the structure of the data. Draft the notice topics to track your Rule 23 elements one-to-one.
  • [ ] Serve targeted written discovery early. Serving Written Discovery Requests: A Practical Checklist and The Federal Discovery Toolkit: A Roadmap From Rule 26(f) to Sanctions cover the mechanics.
  • [ ] Move to compel before your certification deadline, not after. A motion to compel filed two weeks before the brief is due is a motion for a continuance in disguise. See Moving to Compel Discovery and Seeking Rule 37 Sanctions: A Practical Checklist.

Why it matters, and the traps

Certification is an evidentiary motion. You will lose it if you arrive with argument and the defendant arrives with data. Everything in Phases 1 through 9 depends on documents and testimony you must extract on a schedule you must control.

The classic trap: agreeing to "certification-only discovery," then discovering that the court reads that agreement to bar the very substantiation-file and marketing-test discovery that proves your practice was uniform. Define the scope in the discovery plan by topic, not by the label "certification" or "merits."

Phase 11 — Superiority, manageability, and the trial plan

  • [ ] Work the four Rule 23(b)(3) superiority factors on the record: (A) class members' interests in individually controlling separate actions; (B) the extent and nature of litigation already begun by class members; (C) the desirability of concentrating the litigation in this forum; and (D) the likely difficulties in managing a class action.
  • [ ] Lead with the negative-value point where it is true: an individual MetaboClear purchaser has a claim worth perhaps twelve dollars. Without a class, there is no case at all — and that is the strongest superiority argument in the books.
  • [ ] Write an actual trial plan. Not a paragraph: a document. Phase one, common liability, tried to a jury on X, Y, Z evidence in roughly N days. Phase two, damages, computed from the defendant's own data by a claims administrator applying the certified formula.
  • [ ] Address manageability concretely: how members will be identified, how notice will issue (direct email and mail from the sales database, plus a publication and digital-media plan), how claims will be processed, how opt-outs will be handled.
  • [ ] Do not certify what you cannot try. Under Rule 23(e), the standards for a settlement class are not the escape hatch they once were, and Amchem holds that predominance and adequacy still apply in full to settlement classes (manageability alone drops out of the analysis).
  • [ ] For the trial-side mechanics of what you are promising the court, see Federal Civil Trial Practice: From Final Pretrial Conference to Verdict and Preparing for Trial: A Pretrial Order and Trial Readiness Checklist.

Why it matters, and the traps

Judges certify cases they can picture trying. The trial plan is not a formality — it is the exhibit that converts an abstract predominance argument into a concrete, two-week, manageable proceeding. Defense counsel will tell the court that your class is an ungovernable mess. A crisp trial plan is the rebuttal.

The trap is the trial plan that quietly reintroduces individual issues: "at phase two, each class member will testify as to her reliance." That sentence concedes predominance. If your plan needs 40,000 mini-trials, you do not have a class — you have a Comcast and Dukes problem you have written down for the judge.

Phase 12 — Draft the motion and pre-empt the defense

  • [ ] Check the local rules and the judge's standing order for page limits, evidentiary submission format, and any pre-motion conference requirement. Certification briefs commonly get expanded limits — ask.
  • [ ] Structure the brief to the rule: class definition → Rule 23(a) elements → Rule 23(b) category → 23(g) → trial plan → proposed order. Attach the proposed certification order Rule 23(c)(1)(B) requires, including the class definition, the claims and issues certified, and the appointment of class counsel.
  • [ ] Support every factual assertion with a declaration, an exhibit, or deposition testimony. A certification brief with unsupported assertions reads as a complaint.
  • [ ] Pre-empt the five attacks that are coming: (1) fail-safe or overbroad definition; (2) uninjured members under TransUnion; (3) commonality under Dukes; (4) damages-model mismatch under Comcast; (5) atypicality or inadequacy of the representative. Address each affirmatively rather than saving it for reply.
  • [ ] Time the motion under Rule 23(c)(1)(A) — "at an early practicable time" — and against any deadline in the scheduling order or local rule.
  • [ ] Do not treat certification like summary judgment. Your burden is a preponderance on the Rule 23 prerequisites, not the absence of a genuine dispute of material fact. The two motions and their standards are distinct — compare Moving for Summary Judgment Under Rule 56: A Practical Checklist — and the pleading that started it all is governed by different standards still: Drafting a Federal Civil Complaint: A Practical Checklist.
  • [ ] Situate the motion in the arc of the case with The Civil Pretrial Toolkit: From Pleadings to Summary Judgment.

Why it matters, and the traps

The certification brief is the most evidence-dense document most litigators ever write. It is a trial brief, an expert proffer, an evidentiary appendix, and a case-management proposal in one binding. Judges who read a lot of them can tell within ten pages whether the movant has done the work.

The trap is the reply-brief ambush: saving your best answer on uninjured members for the reply, where the court may discount it as sandbagging and the defense gets a sur-reply. Lead with your weaknesses. Own them and solve them.

Phase 13 — Rule 23(f) and the interlocutory posture

  • [ ] Calendar the deadline the day the order issues. A Rule 23(f) petition must be filed with the court of appeals within 14 days after the order granting or denying certification (or, in the government-party circumstance, within 45 days).
  • [ ] Understand that the 14-day deadline is a claim-processing rule not subject to equitable tolling — Nutraceutical Corp. v. Lambert, 586 U.S. 188 (2019). A timely-filed motion for reconsideration can restart the clock in some circumstances; an untimely one cannot. Do not gamble.
  • [ ] Know that review is discretionary. Courts of appeals typically grant where the ruling is a likely "death knell" for the case, where it turns on an unsettled and important legal question, or where it is manifestly erroneous.
  • [ ] Do not try to manufacture an appeal by voluntarily dismissing your claims after a denial and appealing the resulting judgment. Microsoft Corp. v. Baker, 582 U.S. 23 (2017) forecloses that route: such a dismissal does not create appellate jurisdiction over the certification order.
  • [ ] Note that the certification order itself is not otherwise immediately appealable — Coopers & Lybrand v. Livesay, 437 U.S. 463 (1978) rejected collateral-order treatment, and Rule 23(f) exists precisely because of it.
  • [ ] Remember that a Rule 23(f) petition does not stay district-court proceedings unless the district court or the court of appeals so orders. Move for a stay separately if you need one.
  • [ ] For appellate mechanics and standards of review, see Federal Civil Appeals: A Comprehensive Guide and Taking a Civil Appeal: A Practical Checklist.

Why it matters, and the traps

Certification denials kill cases. Rule 23(f) is the only realistic way to fix one before final judgment, and it has a 14-day fuse that has burned down more meritorious appeals than any other deadline in the rules. Put it on the calendar before you read the order.

If certification is granted, expect the defendant's own 23(f) petition — and expect settlement discussions to begin the same week. That is when you turn to Settling a Class Action: A Rule 23(e) Approval Checklist, because a certified class cannot be settled the way an individual case can.

Common mistakes to avoid

  • The fail-safe definition. If a person's membership depends on whether the defendant is liable to her, the class is fail-safe. Rewrite it around objective purchase, employment, or transaction criteria.
  • Defining the named plaintiff out of the class. Redraft, then re-read, then check that your client still qualifies.
  • Ignoring uninjured members until the reply brief. Count them, and pick your answer (narrow, de minimis, or common-question) in Phase 2, not Phase 12.
  • Building the damages model after the theory is locked. Comcast mismatches are created by sequencing, not by economics. The expert and the pleader should be in the same room.
  • Treating Rule 23(g) as boilerplate. Two pages and a declaration. Every time.
  • Certifying a damages class under Rule 23(b)(2). Dukes closed this door. If you want money, you want (b)(3), with its notice costs and opt-out rights.
  • A nationwide class under fifty different state statutes with no variation appendix. Subclass it or lose it.
  • Over-claiming through your expert. Let the expert opine on methodology, not on universal injury. The narrower opinion survives Daubert.
  • A trial plan that requires individual testimony on a predominating element. You have just written the defense's brief.
  • A representative with an arbitration clause, a unique defense, or a credibility problem. Diligence the client before you file, not after the deposition.
  • Missing the Rule 23(f) window. Fourteen days. No tolling. Nutraceutical.
  • Forgetting that certification is not final. Rule 23(c)(1)(C) lets the court alter or amend the order before final judgment — which cuts both ways. Decertification motions are real, and the record you build at certification is the record you will defend for the rest of the case.

The consolidated certification checklist

Class definition

  • [ ] Objective criteria only; no merits or state-of-mind terms
  • [ ] Not fail-safe; membership determinable without adjudicating liability
  • [ ] Named plaintiff falls inside the definition
  • [ ] Class period matches limitations period (plus any tolling)
  • [ ] Standard exclusions (officers, directors, judge, opt-outs) included
  • [ ] Subclasses proposed where products, states, or statutes diverge
  • [ ] Circuit's ascertainability standard identified and satisfied

Standing and injury

  • [ ] Every member has concrete Article III injury (TransUnion)
  • [ ] Uninjured-member share quantified and answered
  • [ ] Injury provable as a common question, or definition narrowed

Rule 23(a)

  • [ ] Numerosity proved with data, not estimate
  • [ ] Commonality framed as a common answer from common evidence (Dukes)
  • [ ] The "glue" — uniform policy, label, contract, or algorithm — in the record
  • [ ] Typicality: no unique defenses, no arbitration mismatch
  • [ ] Adequacy: no conflicts, credible representative, engaged client

Rule 23(g)

  • [ ] All four 23(g)(1)(A) factors addressed by declaration
  • [ ] Leadership and fee arrangements resolved and disclosed

Rule 23(b)

  • [ ] Correct category selected; no damages under (b)(2)
  • [ ] Predominance matrix complete, element by element
  • [ ] Reliance addressed (objective standard or Basic presumption)
  • [ ] State-law variation appendix if multistate
  • [ ] Superiority: negative-value point made; four factors briefed
  • [ ] Rule 23(c)(4) issue class pleaded in the alternative

Damages

  • [ ] Model measures only the certified theory (Comcast)
  • [ ] Model runs on defendant's own data; demonstrated, not promised
  • [ ] Individualized amounts distinguished from unworkable methodology

Expert

  • [ ] Rule 702 (as amended December 2023) satisfied on every essential opinion
  • [ ] Complete Rule 26(a)(2)(B) report, data, and code produced
  • [ ] Own expert stress-tested against a full Daubert attack
  • [ ] Rebuttal expert retained if needed

Discovery

  • [ ] Certification-discovery schedule negotiated at the Rule 26(f) conference
  • [ ] Hard merits/certification bifurcation resisted
  • [ ] Transaction database and data dictionary obtained
  • [ ] Rule 30(b)(6) deposition taken on uniformity and data
  • [ ] Motions to compel filed with time to spare

Motion

  • [ ] Local rules and standing order checked (page limits, pre-motion conference)
  • [ ] Filed "at an early practicable time" (Rule 23(c)(1)(A))
  • [ ] Every fact supported by declaration, exhibit, or transcript
  • [ ] Trial plan attached
  • [ ] Proposed Rule 23(c)(1)(B) order attached
  • [ ] The five standard defense attacks pre-empted affirmatively

After the order

  • [ ] Rule 23(f) 14-day deadline calendared the day the order issues
  • [ ] Stay motion prepared if needed
  • [ ] Rule 23(c)(2)(B) notice plan ready if certification is granted

Primary Authority

Rules

Supreme Court

Courts of appeals (note the splits)

Secondary sources

  • Newberg and Rubenstein on Class Actions (the standard treatise; consult the current edition)
  • Manual for Complex Litigation (Fourth), Federal Judicial Center — §§ 21.1–21.6 on class-action management
  • Wright & Miller, Federal Practice and Procedure §§ 1751–1805 (class actions)

Related UpLaw Resources

Related Guides & Articles

Related Checklists

Related Toolkits

This checklist is general legal information from UpLaw (uplaw.us), not legal advice, and does not create an attorney-client relationship. Class-certification law varies by circuit and changes rapidly — particularly on ascertainability, the treatment of uninjured class members, and the standard for expert evidence at certification. Verify current requirements in your forum and consult qualified counsel before acting.