Trademark Survey Design and Challenge Checklist: Universe, Controls, and Admissibility

By Casey Scott McKay · 2025-04-06

This checklist runs a trademark survey program from the first "should we survey at all" memo through the motion to exclude, in twelve phases a litigator can work top to bottom. It covers the go/no-go decision and its written record, the six-month calendar you must build backward from the Rule 26(a)(2)(D) disclosure date, retaining the expert under Rule 26(b)(4)(D) so the exploratory phase stays protected, drafting a screener that operationalizes the legal universe, building a stimulus that survives a realism attack under Sazerac, designing a one-variable control, piloting at n=50-75, and locking a signed, dated protocol before the first respondent is fielded. It then covers the paperwork: the six elements of a Rule 26(a)(2)(B) report, the twelve-item backup production list with file formats, respondent anonymity, and privilege-log discipline for the three Rule 26(b)(4)(C) carve-outs. The back half is adversarial — reanalyzing the opponent's respondent-level data, a ten-topic deposition plan, the three grounds that actually win a motion to exclude under the December 2023 amendment to Federal Rule of Evidence 702, and the opposition playbook when you are defending. It closes with the TTAB variant and its separate thirty-day expert-disclosure clock, the compressed preliminary-injunction schedule, a common-mistakes list, and a deadlines table. A single worked matter, Kestrel Athletic against Kestral Fit, runs through every phase so you can see what "done" looks like.

IP and Technology > Trademarks | Checklist | Published 6 April 2025 - Updated 7 April 2026 | Casey Scott McKay - marksy.us

Summary. Twelve phases, worked top to bottom, that take a survey program from the go/no-go memo to the motion to exclude. You get the backward calendar from the Rule 26(a)(2)(D) date, the Rule 26(b)(4)(D) retention structure that keeps exploratory work protected, a screener that implements the legal universe, a stimulus built from produced documents rather than screenshots, a one-variable control, a pilot, a signed pre-fielding protocol, the six elements of the Rule 26(a)(2)(B) report, a twelve-item backup production list with file formats, a ten-topic deposition plan, and the three grounds that actually win a motion to exclude after the December 2023 amendment to Federal Rule of Evidence 702. TTAB and emergency-injunction variants are handled separately, because the clock and the payoff both change. One invented matter, Kestrel Athletic v. Kestral Fit, runs through every phase.

Keywords: trademark survey checklist · survey universe screener · control cell design · eveready survey · squirt survey · net confusion · daubert motion to exclude survey · federal rule of evidence 702 · rule 26(a)(2)(b) expert report · survey backup data production · deposing a survey expert · survey protocol document · stimulus realism · ttab expert disclosure · 37 cfr 2.120 · online panel data quality · pilot survey · verbatim coding codebook · preliminary injunction survey declaration · rebuttal survey expert

What this checklist is for

This is the working document for anyone who has to approve a trademark survey protocol, defend one, or take one apart. It assumes you already know what the formats measure. If you cannot say in one sentence why an Eveready result of 2% proves nothing about a regional mark, stop and read Consumer Surveys in Trademark Cases: What Eveready and Squirt Actually Measure first. The reasoning behind each box below — why a control is non-negotiable, why counsel must not draft the questionnaire, why a denied Daubert motion is worse than no motion — is in Commissioning and Attacking a Trademark Survey: A Litigator's Guide to Design, Universe, and Daubert. This document does not re-teach any of it. It tells you what to do, in order.

Who should use it. Litigators on either side of a federal trademark case; in-house counsel approving a six-figure expert budget; TTAB practitioners deciding whether a survey is rational in a single-class opposition; and anyone just served with the other side's expert report who has thirty days.

What you'll need before you start. The operative complaint and the pleaded confusion theory (forward, reverse, post-sale, initial-interest — they take different surveys); the Rule 16(b) scheduling order with the Rule 26(a)(2)(D) dates; revenue and awareness data for the senior mark; the accused party's marketplace listings, packaging, and hang tags, produced under Fed. R. Civ. P. 34 rather than screenshotted; written budget authority; and a copy of Shari Seidman Diamond, Reference Guide on Survey Research, in Reference Manual on Scientific Evidence 359 (Fed. Jud. Ctr., 3d ed. 2011), which is the work your judge will consult.

The worked matter. Kestrel Athletic, Inc. — a Portland running-shoe company, $640 million in annual revenue, 78% aided awareness among adults who bought athletic footwear in the last twelve months, registered word mark KESTREL in Class 25 — sued Kestral Fit LLC in the District of Oregon on 12 March 2025 under 15 U.S.C. § 1114(1) and § 1125(a) over KESTRAL FIT leggings sold direct-to-consumer and on two marketplaces. We follow Kestrel's survey from retention to trial, and attack Kestral Fit's.

| Phase | What you accomplish | Typical elapsed time | |---|---|---| | 1 | Decide whether to survey, and document the decision | 1 week | | 2 | Build the calendar backward from the disclosure date | 1 day | | 3 | Vet and retain the expert under Rule 26(b)(4)(D) | 2 weeks | | 4 | Define the universe; write the screener | 2 weeks | | 5 | Build the stimulus from produced documents | 2 weeks | | 6 | Design the control and the questionnaire | 2 weeks | | 7 | Pilot at n = 50-75; lock and sign the protocol | 3 weeks | | 8 | Field, code, tabulate, preserve | 6-8 weeks | | 9 | Serve the Rule 26(a)(2)(B) report; produce the backup | 4 weeks | | 10 | Reanalyze and depose the opposing expert | 3-5 weeks | | 11 | File or oppose the motion to exclude | 3 weeks | | 12 | Forum variants: TTAB, and the emergency schedule | as applicable |

Phase 1 — Decide whether to survey, and prove you decided

[ ] Write the legal question in one sentence, in the language of the statute or the circuit's factor test, and put it in a memo to the file.
- Why. Whether an appreciable number of ordinarily prudent prospective purchasers of women's athletic apparel believe KESTRAL FIT goods come from, or are affiliated with or approved by, the source of KESTREL. If you cannot write that sentence, no questionnaire will save you.
- Authority. 15 U.S.C. § 1114(1), § 1125(a); Mushroom Makers, Inc. v. R.G. Barry Corp., 580 F.2d 44, 47 (2d Cir. 1978) (appreciable number of ordinarily prudent purchasers).
[ ] Confirm the pleaded confusion theory matches the survey you are about to buy — forward, reverse, post-sale, or initial-interest.
- Trap. A competent point-of-sale survey proves nothing in a case pleading only post-sale confusion. The number can be perfect and still irrelevant.
[ ] Verify the perception you need is a present one, not a historical one.
- Authority. Pro-Football, Inc. v. Harjo, 284 F. Supp. 2d 96, 132 (D.D.C. 2003) (1996 survey irrelevant to 1967-1990 perceptions). Retrospective fame under 15 U.S.C. § 1125(c)(2)(A) and secondary meaning as of a priority date years back are the recurring casualties.
[ ] Get an incidence estimate from a panel provider before quoting a number to the client.
- Why. At 22% incidence you screen roughly 3,600 people to complete 800. At 2% you screen 40,000, and the sample line moves from $12,000 to $70,000.
[ ] Present the client a written budget for the program, not the fielding: design and report $35,000-$85,000; programming $3,000-$8,000; consumer sample at n = 800 $8,000-$25,000; pilot $2,000-$6,000; blind coding $4,000-$12,000; deposition $10,000-$30,000; rebuttal expert $25,000-$60,000 (critique) or $60,000-$140,000 (counter-survey); motion practice $25,000-$60,000. All-in, one side, through trial: $120,000-$300,000.
- Trap. Clients hear "$60,000" and budget $60,000. A survey that matters costs roughly the same again to defend, and expert fees are generally not recoverable even in an exceptional case under 15 U.S.C. § 1117(a). Say so in the engagement email. See What a Trademark Win Is Worth.
[ ] Run the no-survey test: are the marks effectively identical, the goods identical, and the intent record bad? If yes, do not survey.
- Why. You will win on the factors described in Trademark Infringement: Proving Likelihood of Confusion. A survey adds cost, delay, and a target.
[ ] Where you decide not to survey, fund the substitutes and say so in the memo: a logged actual-confusion intake (misdirected calls, wrong-company returns, retailer emails, marketplace reviews), marketplace co-occurrence screenshots, the defendant's naming files and ignored clearance search, and a clean chain of title.
- Authority. No adverse inference in most circuits. Charles Jacquin et Cie, Inc. v. Destileria Serralles, Inc., 921 F.2d 467 (3d Cir. 1990); Le Sportsac, Inc. v. K Mart Corp., 754 F.2d 71 (2d Cir. 1985). A minority disagree. Braun Inc. v. Dynamics Corp. of Am., 975 F.2d 815 (Fed. Cir. 1992).
[ ] Date and sign the memo. It is your answer to the cross-examination that you shopped for a format until you found a number.

Kestrel, Phase 1. Counsel's 2 April 2025 memo concludes: forward confusion; senior mark commercially strong; goods related but not shelf-adjacent; therefore aided Eveready, not Squirt. Budget approved at $180,000 with a $60,000 contingency for a rebuttal expert.

Phase 2 — Build the calendar backward from the disclosure date

[ ] Pull the Rule 16(b) scheduling order and calendar the opening expert disclosure date, the rebuttal date, the expert-discovery cutoff, and the Daubert filing date.
- Authority. Fed. R. Civ. P. 26(a)(2)(D)(i) (absent a stipulation or order, 90 days before trial); Fed. R. Civ. P. 26(a)(2)(D)(ii) (rebuttal within 30 days of the other side's disclosure); Fed. R. Civ. P. 26(e)(2) (supplementation due when Rule 26(a)(3) pretrial disclosures are due, i.e., 30 days before trial).
[ ] Subtract six months from the opening disclosure date and mark that as the last honest retention date.
- Why. Design 5 weeks, pilot 2, lock and program 2, field 3-4, code and tabulate 4, report 4. There is no compression that does not show.
[ ] If you have fewer than four months, put the choice to the client in writing: a survey that will not survive a deposition, or no survey.
- Trap. The single most common survey failure is arithmetic, not methodology.
[ ] Serve Fed. R. Civ. P. 34 requests now for the accused party's packaging, hang tags, labels, native marketplace listings, product photography, keyword buys, and channel data — you need them for the stimulus in Phase 5, and produced documents authenticate themselves in a way screenshots do not.
[ ] Preserve marketplace evidence the same week: dated, full-URL captures of the accused listings at desktop and mobile resolution, and the first-page results for the three or four queries a real shopper would run.
- Trap. Defendants redesign. A stimulus built from an archived page invites "that is not what consumers see today."

Phase 3 — Vet and retain the expert so Rule 26(b)(4) actually protects you

[ ] Pull every published decision discussing your candidate and read for a habit — a signature universe definition, a house control, a favored question wording — that is about to reappear in your case.
[ ] Ask for the Rule 26(a)(2)(B)(v) list (four years of trial and deposition testimony) before you sign, and read it for a party-side pattern.
[ ] Retain in a non-testifying consulting capacity for the exploratory phase and convert in writing later.
- Authority. Fed. R. Civ. P. 26(b)(4)(D) (consulting experts generally undiscoverable); Fed. R. Civ. P. 26(b)(4)(B) (draft reports protected "regardless of the form in which the draft is recorded").
- Trap. Conversion is not a laundry. Rule 26(a)(2)(B)(ii) requires disclosure of facts or data "considered," read broadly to include material reviewed, generated, or rejected. Republic of Ecuador v. Mackay, 742 F.3d 860, 870 (9th Cir. 2014). If the pilot shaped the fielded design, plan on producing it.
[ ] Write the retention letter with these eight numbered paragraphs: scope and capacity; expert independence over design; a bare-transmittal rule for facts; a single numbered "Assumptions Provided by Counsel" exhibit; native-format preservation; blinding of every vendor, coder, and respondent; non-contingent compensation; and written designation as the trigger for Rule 26(a)(2)(B).
- Why. Paragraph four is the one people skip and the one that pays. At deposition, "what did the lawyers tell you to assume" is answered by handing over a one-page exhibit you drafted to be read aloud.
[ ] Separate your theory of the case from your transmittals.
- Authority. Fed. R. Civ. P. 26(b)(4)(C)(i)-(iii) makes discoverable only compensation, facts or data you supplied that the expert considered, and assumptions you supplied that the expert relied on. Communications about the relevance of those facts stay protected. Republic of Ecuador v. Hinchee, 741 F.3d 1185, 1190-91 (11th Cir. 2013).
[ ] If the case is important enough that you may need to bury a bad study, retain two people: a consultant who runs exploratory work and a testifying expert who never sees it.
[ ] Decline any expert who predicts a favorable number before seeing the marks and the category.
- Trap. The correct answer to "will this help us?" is "I don't know until I run a pretest." An expert who promises results has told you how they design studies.

Phase 4 — Define the universe and write the screener

[ ] State the legally relevant population in one line, sourced to the claim: forward confusion → the defendant's prospective purchasers; reverse confusion → the plaintiff's; secondary meaning → prospective purchasers of the plaintiff's goods; genericness → the relevant purchasing public for that specific category; dilution fame → the general consuming public of the United States.
- Authority. 15 U.S.C. § 1125(c)(2)(A) (fame); 15 U.S.C. § 1064(3), § 1127 (primary significance).
- Trap. Getting forward/reverse backwards is a total loss, not a discount. Several courts treat universe error as an irrelevance problem rather than a weight problem.
[ ] Screen to prospective purchasers — past purchasers within the category's real cycle or near-term intenders — never to current customers of either party.
- Why. Screening to the plaintiff's customers primes them to see the plaintiff everywhere; screening to the defendant's buyers measures post-purchase regret.
- Authority. Amstar Corp. v. Domino's Pizza, Inc., 615 F.2d 252 (5th Cir. 1980) (faulting a survey of women at home about pizza in markets where the defendant did not operate).
[ ] Delete any screener item that asks whether the respondent has heard of the senior mark.
- Trap. This is the single most reliable cross-examination hit in the field. It builds the conclusion into the sample.
[ ] Match the geographic screen to the scope of the rights asserted.
- Why. A national panel in a case about regional common-law rights is a mismatch. See Where Your Trademark Rights End and Establishing and Proving Common-Law Trademark Rights.
[ ] Set the purchase-recency window to the category's actual cycle: 12 months for apparel, 3 months for coffee, 7 years for mattresses.
[ ] Include security screens that terminate anyone in market research, advertising, the parties' industry, or a law firm, and a household-member version of each.
[ ] Hide the subject inside a rotated multi-select category list with at least four unrelated distractors.
[ ] Record device type and screen size without terminating on it.
- Why. When the other side argues your stimulus was unviewable on a phone, you want the device mix in the data, not in your memory.
[ ] Produce the screener as a standalone exhibit with the termination logic visible as a flowchart.

Kestrel, Phase 4. S3 asks what the respondent bought in the past 12 months from a rotated list (kitchen appliances / women's athletic apparel such as leggings, sports bras, or workout tops / lawn and garden / pet supplies / luggage); S4 asks the same about the next 12 months; qualifying on either. Nothing in the screener names either party. Incidence came back at 24%; the sample line was priced at $14,200 for n = 800 across two cells.

Phase 5 — Build the stimulus

[ ] Build every image from documents produced under Fed. R. Civ. P. 34, not from a paralegal's screenshot.
- Why. A produced document is authenticated. A screenshot is a foundation fight you will have at the worst moment.
[ ] Show the mark as it is actually used — house brand, color scheme, price, marketplace chrome, and all — at the resolution and on the devices consumers actually see.
- Authority. Sazerac Co. v. Fetzer Vineyards, Inc., 265 F. Supp. 3d 1013 (N.D. Cal. 2017) (69% result discarded because the shelf set was one no store would assemble).
[ ] In trade dress cases, show the claimed dress as claimed, stripped of the word marks that ordinarily do the identifying work.
- Trap. Uncomfortable and unavoidable. See Protecting Trade Dress and the Trade Dress Protection Checklist. For sound, motion, and scent marks the stimulus decision is itself substantive — see Color, Sound, Scent, and Motion.
[ ] Where the design calls for an array, populate it with the accused party's actual first-page competitors, captured on a stated date from a stated query, and write down why each member is there.
[ ] Randomize array position, array order, and question-block order, and store the rotation assignment in the respondent-level data.
- Why. An expert who cannot say whether position moved the result has left a hole a rebuttal expert will find.
[ ] Never enlarge, center, isolate, crop, or clean up the accused mark.
- Trap. Every enhancement becomes a deposition exhibit, and each one is easy for a judge to see. If the real listing shows the mark at 14 pixels among forty thumbnails, that is your stimulus.
[ ] Write a dated design memo listing each stimulus file, its provenance, its pixel dimensions, and its presentation conditions.

Kestrel, Phase 5. Room one: a simulated category page, twelve women's athletic apparel listings including KESTREL and eight competitors that were the actual top results for "women's leggings" and "workout leggings" on both marketplaces on 4 April 2025. Distractor: a two-minute word game. Room two: the KESTRAL FIT product detail page alone, as captured, price and chrome intact. That one paragraph is the answer to Sazerac.

Phase 6 — Design the control and write the questions

[ ] Name the single variable the control isolates, in writing, before anyone drafts a control stimulus.
- Authority. Diamond, Reference Guide on Survey Research, in Reference Manual on Scientific Evidence 359 (Fed. Jud. Ctr., 3d ed. 2011): the control should share as many characteristics with the test stimulus as possible except the characteristic being measured.
[ ] Hold everything else constant — same photography, same layout, same price, same chrome, same room-one array — and change only the mark.
[ ] Invent the control mark, then clear it against the USPTO register and a web search.
- Trap. A control that turns out to be somebody's real brand imports that brand's equity and produces a bad afternoon.
[ ] Reject a control that is too far from the test mark.
- Why. In a suit between 24 HOUR FITNESS and 24/7 FITNESS, LIFETIME FITNESS as a control cannot separate respondents who linked the parties because the marks are similar from respondents who linked them because both gyms advertise being open around the clock. THE 24 HOUR GYM does that work.
[ ] State the cell allocation and the arithmetic before fielding: n = 400 per cell yields roughly ±5 points per cell at 95% confidence and a net-confusion interval of roughly ±7 points.
- Trap. If your predicted net is 12 points and your interval is ±7, say so in the protocol. Discovering it in the report is worse.
[ ] Refuse to field without a control, whatever the budget says.
- Why. Raw numbers are uninterpretable. Water Pik, Inc. v. Med-Systems, Inc., 726 F.3d 1136 (10th Cir. 2013) (23.15% test, 16.7% control, net under 6.5%, "insignificant"); 1-800 Contacts, Inc. v. Lens.com, Inc., 722 F.3d 1229 (10th Cir. 2013) (19.3% minus 11.9% = 7.4%, too low); compare Combe Inc. v. Dr. August Wolff GmbH & Co., 382 F. Supp. 3d 429 (E.D. Va. 2019) (37% minus 18% = 19 points, "reliable and powerful").
[ ] Offer "don't know" on screen and instruct respondents not to guess.
- Why. A forced two-option choice on a coin-flip question manufactures roughly half a point of fake confusion for every guesser.
[ ] Include a "why do you say that?" probe after every substantive answer, and rotate answer order.
- Trap. Without the probe you cannot separate a respondent confused by the marks from one confused by the category, the packaging color, or the retailer. "Because all flashlight bulbs are Eveready" is a generic answer, not a confused one.
[ ] Delete any question that names both parties, pairs them, or supplies the theory of connection.
- Authority. Simon Prop. Grp. L.P. v. mySimon, Inc., 104 F. Supp. 2d 1033 (S.D. Ind. 2000) (excluding proposed surveys for suggestive questions and absent controls); THOIP v. Walt Disney Co., 690 F. Supp. 2d 218, 235 (S.D.N.Y. 2010) (sequential presentation proper "only if it reflects a significant number of real world situations" in which both marks are evaluated together).

Phase 7 — Pilot, then lock and sign the protocol

[ ] Field 50 to 75 respondents on the locked draft and check six things: screener incidence against budget, comprehension of the open ends, control cell behavior, speeder and straightliner rate, codeability of verbatims, and median length of interview.
- Why. $3,000 to find out whether you are about to waste $60,000. Cheapest insurance in the program.
[ ] If the control returns anything near double digits, rebuild it — it shares the variable you meant to remove.
[ ] If the instrument is broken, revise and re-pilot. Do not field a broken instrument on the theory that a larger n will fix it.
[ ] If the instrument works but the signal is flat, stop and reassess the case theory while your expert is still a consultant.
- Trap. The version that gets people in trouble is the "pilot" that is a full study with a bad number, quietly re-run until it comes out right. It is discoverable more often than people think, and it converts a methodology fight into a credibility fight you cannot win.
[ ] Assemble the protocol with all twelve components: hypothesis and legal question; format selected and formats rejected, with reasons; universe and screener with termination logic; sampling plan with quotas and expected margin of error; stimulus specification with provenance and rotation scheme; control specification and the one-variable justification; the full respondent-facing questionnaire with skip logic; the blinding protocol; the data-quality plan; the coding plan; the analysis plan; and the preservation plan.
[ ] Fix the data-quality removal rules in advance: attention-check pass criteria, a speeder threshold (commonly one-third of median length of interview), straightliner detection, duplicate-device and duplicate-IP screening, and gibberish or copy-paste detection in open ends.
- Trap. A speeder threshold chosen after you see which cell the speeders are in is not a quality control. A good rebuttal expert will reconstruct the date the rule was adopted from the data files.
[ ] Have the expert sign and date the protocol before the first respondent is fielded, and attach it to the report as Appendix A.
- Why. It converts a hundred deposition questions into one answer: that decision was made on 12 June, before I saw any data, and here is the document.

Phase 8 — Field, code, tabulate, and preserve

[ ] Blind the panel provider, the programmer, the interviewers, the coders, and the respondents to the sponsor, counsel, and the existence of litigation.
- Authority. The seven traditional foundation elements include competent interviewers with no knowledge of the litigation or its purpose; the absence of blinding is a reliable cross-examination theme.
[ ] Override any vendor default that purges data after 90 days, in the contract.
[ ] Preserve in native format: respondent-level data with all timestamps, raw open-end text, the AAPOR-style disposition report (completes, terminates by screener item, quota-full, abandons), the programming file, every instrument version, all vendor correspondence, and the panel provider's written description of its sample source.
[ ] Do not monitor results by cell while fielding is open.
- Trap. If your expert peeks and then adjusts anything, the timestamps will say so.
[ ] Build the codebook from a sample of verbatims before coding begins; double-code a stated minimum share; report an intercoder agreement statistic; and document the adjudication procedure for disagreements. Coders stay blind to cell and sponsor.
[ ] Compute net confusion as test cell minus control cell, and report the confidence interval alongside it.
[ ] Report the data-quality removals step by step, by cell, with the date each rule was set.

Phase 9 — Serve the Rule 26(a)(2)(B) report and produce the backup

[ ] Confirm the report contains all six required elements: (i) a complete statement of all opinions with basis and reasons; (ii) the facts or data considered; (iii) exhibits used to summarize or support; (iv) qualifications with ten years of publications; (v) four years of prior trial and deposition testimony; (vi) compensation.
- Authority. Fed. R. Civ. P. 26(a)(2)(B). "Expert reports must not be sketchy, vague or preliminary in nature." Salgado v. Gen. Motors Corp., 150 F.3d 735, 741 n.6 (7th Cir. 1998).
[ ] Include a Limitations section that states what the study does not show.
- Why. After the December 2023 amendment adding Fed. R. Evid. 702(d) — the opinion must "reflect a reliable application of the principles and methods to the facts of the case" — a conclusion that outruns the data is an admissibility problem, not merely a rhetorical one.
[ ] Do not draft or heavily edit the report.
- Authority. United States ex rel. Wall v. Vista Hospice Care, Inc., 319 F.R.D. 498, 509-10 (N.D. Tex. 2016); Numatics, Inc. v. Balluff, Inc., 66 F. Supp. 3d 934, 943-44 (E.D. Mich. 2014). You may tell the expert to write for an intelligent lay reader and to confront the other side's best point. That is the line.
[ ] Produce the complete backup on the day of the report — twelve items, in these formats: signed protocol (PDF); screener and questionnaire (PDF plus the native programming file); stimulus files (native images and video at fielded resolution, every rotation variant); respondent-level data (.SAV or CSV with a labeled data dictionary, not a PDF of crosstabs); raw verbatims keyed to respondent ID; coded verbatims with codebook, coder ID, and the double-coded subset; AAPOR-style disposition report; the data-quality log; field vendor contracts, instructions, and invoices; the panel provenance statement; tabulations with syntax; and the Rule 26(a)(2)(B)(ii) materials-considered list with copies.
- Trap. The most effective attack on a survey is not a design critique. It is "we asked for the underlying data and got a PDF of tables." Courts have discounted or excluded survey testimony where the proponent did not disclose enough to permit verification.
[ ] Redact respondent names, emails, resolvable panel IDs, and IP addresses while preserving a stable pseudonymous respondent ID.
- Why. Research-ethics codes require it, and done this way it is uncontroversial and forecloses any attempt to depose respondents. Testimony from the expert who designed and supervised the study lays the foundation; the interviewers need not be deposed. See Piper Aircraft Corp. v. Wag-Aero, Inc., 741 F.2d 925 (7th Cir. 1984).
[ ] Log withheld drafts and communications by category, and release the three carve-outs without a fight.
- Authority. Fed. R. Civ. P. 26(b)(4)(B), (C). Some courts read "draft report" narrowly and order production of notes, task lists, and outlines. In re Application of Republic of Ecuador, 280 F.R.D. 506, 513 (N.D. Cal. 2012), aff'd sub nom. Republic of Ecuador v. Mackay, 742 F.3d 860 (9th Cir. 2014). Tell your expert to write as though the notes will be produced.
[ ] Diary the Rule 26(a)(2)(D)(ii) rebuttal date the moment you are served, and the Rule 26(e)(2) supplementation date at 30 days before trial.
- Trap. Fed. R. Civ. P. 37(c)(1) supplies exclusion as the remedy for a late disclosure unless it was substantially justified or harmless. Surveys run late; this is a live risk, not a theoretical one.

Phase 10 — Take the opposing survey apart

[ ] Take the survey yourself, from a live test link or a rebuild from the programming file.
- Why. You will find an image that renders at 400 pixels on a phone, a forced-response question with no "don't know," or a rotation flag that never rotated. Reading the report finds none of that.
[ ] Re-run the topline from the respondent-level data and reconcile it against the report.
- Why. Discrepancies between the data and the report are the highest-value finding available and more common than you would like.
[ ] Recode a sample of verbatims to a defensible codebook and report the resulting net.
- Why. If 23 points become 11 under a coding rule any competent researcher would accept, you have a demonstrative no methodological argument can match.
[ ] Recompute the net with and without each discretionary data-quality removal.
[ ] Build a one-page protocol tree: every design decision, the alternative rejected, and the date the choice was made. That is your deposition outline.
[ ] Cover ten topics in the deposition: scope and independence; format selection; universe; sample and administration; stimulus; questions and order; control; coding and verbatims; data quality; and opinion scope under Rule 702(d).
- Authority. Fed. R. Civ. P. 30(d)(1) (seven hours); Fed. R. Civ. P. 26(b)(4)(E) (you pay the expert's reasonable fee for time spent responding).
[ ] Ask the "what would it mean if" questions in the abstract before you show the expert the flaw.
- Why. Lock the answer while the witness still does not know why you asked.
[ ] Check the scheduling order before you teach the expert what is wrong with the study.
- Trap. If a corrected or supplemental report is still permitted, a thorough demonstration at deposition buys the other side a fix. Take the concessions that cannot be un-said and save the demonstration for the motion.
[ ] Decide critique-only versus counter-survey on the merits, not the calendar: critique when the defect is structural (wrong universe, no control, manufactured stimulus); counter-survey only when you have 10-14 weeks, which you almost never do inside the 30-day rebuttal window.
- Trap. A "rebuttal" expert who fields a new affirmative study inside Rule 26(a)(2)(D)(ii)'s window is not rebutting. That draws a Rule 37(c)(1) motion you will probably lose.

Phase 11 — File the motion to exclude, or oppose it

[ ] Ask first whether this is a jury trial. If it is a bench trial, file only for a structural defect that could produce actual exclusion; otherwise put the critique in the trial brief.
- Why. "Flaws go to weight" is worth a great deal in front of a jury and almost nothing in front of a judge — the same person denies your motion and then gives the survey zero weight in the findings.
[ ] Confirm you have at least one of the three grounds with a real exclusion record: wrong universe, no control, or manufactured stimulus. If you have none, file a narrow Rule 702(d) motion to limit the opinion, or nothing.
- Trap. A denied Daubert motion is not neutral. It produces a written order the other side quotes in opening and in every subsequent brief.
[ ] Lead the introduction with the single defect stated as a fact, with a record cite: "Dr. X surveyed purchasers of Kestrel's shoes. The legally relevant population in a forward-confusion case is purchasers of the accused goods. He surveyed the wrong people."
[ ] Plead the amended standard accurately: the proponent must demonstrate to the court that it is more likely than not that each Rule 702 condition is met, and the opinion must reflect a reliable application of the method to the facts.
- Authority. Fed. R. Evid. 702 (amended Dec. 1, 2023); Fed. R. Evid. 104(a); Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993); Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
[ ] Argue two grounds, not six, and add Fed. R. Evid. 403 as the fallback where circuit law says flaws go to weight.
[ ] Attach the exhibits. The stimulus-realism argument is won by putting the real listing beside the survey image, not by adjectives.
[ ] Do not lead with hearsay.
- Authority. Zippo Mfg. Co. v. Rogers Imports, Inc., 216 F. Supp. 670 (S.D.N.Y. 1963); Fed. R. Evid. 703, 803(1), 803(3), 807; Schering Corp. v. Pfizer Inc., 189 F.3d 218, 227-30 (2d Cir. 1999). Reserve it for a survey built on recollection of long-past events.
[ ] When opposing, lead with the majority rule, then meet the 2023 amendment item by item with record cites, then attach the signed pre-fielding protocol.
- Authority. Schering, 189 F.3d at 228; Fortune Dynamic, Inc. v. Victoria's Secret Stores Brand Mgmt., Inc., 618 F.3d 1025, 1036-38 (9th Cir. 2010); Southland Sod Farms v. Stover Seed Co., 108 F.3d 1134, 1143 n.8 (9th Cir. 1997).
- Trap. An opposition that pretends the amendment did not happen will be read by a judge who has just read the advisory committee's warning that many courts were too permissive.
[ ] Comply with the local meet-and-confer requirement and the judge's individual rules on separate memoranda and page limits before filing.

Phase 12 — Forum variants: the TTAB and the emergency schedule

[ ] At the Board, calendar the expert disclosure to thirty days before the close of discovery, not to the close of discovery.
- Authority. 37 C.F.R. § 2.120(a)(2)(iii); TBMP § 401.03. Miss it and your survey may never be considered; make it, and the Board will typically permit the other side to take discovery of your expert and may suspend or extend proceedings to allow it.
[ ] Plan to get the report in through testimony, not a notice of reliance.
- Authority. A survey report is not a printed publication under 37 C.F.R. § 2.122(e). It enters by testimony deposition or by affidavit or declaration under 37 C.F.R. § 2.123(a), subject to the adverse party's right to elect oral cross-examination. Pretrial disclosures are due no later than fifteen days before the opening of the testimony period. 37 C.F.R. § 2.121(e).
[ ] Skip the Daubert-style motion to strike in a Board case and budget a rebuttal declarant plus cross-examination instead.
- Why. No jury, no gatekeeping payoff; the Board admits and weighs. Also budget for the paper: the Board never sees the witness, so internal consistency outranks presentation.
[ ] Run the cost test before recommending a Board survey: a single-class opposition competently tried might cost $25,000-$60,000; a defensible confusion survey with rebuttal and cross adds $60,000-$120,000; there is no fee shifting and no damages, only registrability. Check the current ESTTA filing fees at 37 C.F.R. § 2.6 before quoting anything to the client.
- Trap. Spending twice the cost of the proceeding to block one registration is almost never rational when the objective is coexistence. See Federal Court vs. TTAB: Where to Bring Your Dispute, Understanding TTAB Discovery and the Protective Order, the TTAB Opposition Filing Checklist, and the Trademark Coexistence Agreement — Template.
[ ] Make the exception for genericness and acquired distinctiveness, where a survey is often the only evidence that speaks to the statutory question.
- Authority. 15 U.S.C. § 1064(3), § 1127 (primary significance); 15 U.S.C. § 1052(f) and TMEP § 1212.06(d) (survey evidence of acquired distinctiveness); 37 C.F.R. § 2.41(a)(2) (five years of substantially exclusive use is a rebuttable shortcut, not proof). See Claiming Acquired Distinctiveness at the USPTO, the Secondary Meaning Evidence Checklist, the Genericness Defense and Prevention Checklist, and How to Overcome a Descriptiveness §2(e)(1) Refusal.
[ ] On a TRO or preliminary-injunction schedule, compress to four weeks — week 1 retain and collect; week 2 instrument, control, program, micro-pilot at n = 30; week 3 field n = 300-400 across two cells; week 4 code, tabulate, and sign an expert declaration under Fed. R. Civ. P. 65 practice — and keep the control.
- Why. A two-cell study at n = 200 per cell is defensible. A one-cell study at n = 400 is not.
- Trap. The emergency record follows you. If the expedited survey returns 9% net and the trial study returns 24%, the defendant will put both in front of the same judge. See Moving for a TRO or Preliminary Injunction in a Trademark Case and the Preliminary Injunction Motion Checklist.

Common Mistakes

Deciding to survey eight weeks before the expert deadline. Everything downstream is compromised, and the compromises are visible in the protocol dates.
Retaining the testifying expert first. No consulting phase means no protected exploratory work and no cheap way to learn the answer is bad.
Screening on awareness of the senior mark. The conclusion is now in the sample, and every competent opponent knows to look for it.
Surveying the plaintiff's customers in a forward-confusion case. Wrong universe is the one defect that draws exclusion rather than discount.
Fielding with no control because the budget was halved. A raw 23% is uninterpretable; Water Pik and Combe are the same arithmetic with opposite outcomes.
Building the stimulus from a screenshot. You inherit a foundation fight and a "that is not what consumers see" cross-examination.
Setting speeder and straightliner rules after seeing results by cell. The data files preserve the order of operations.
Letting counsel draft the report. A ghostwritten report is a sanctions exposure and a fatal credibility problem.
Producing tables instead of respondent-level data. The fight about hiding the ball is worse than anything in the data.
Running the survey the format fits rather than the one the pleaded theory requires. A point-of-sale study in a post-sale case measures the wrong moment.
Filing a six-ground motion to exclude in a bench trial. You buy an order calling the other side's survey reliable.
Treating a low Eveready number as proof of no confusion when the senior mark is weak. It disproves unaided recall, which the mark could never have passed.

Deadlines at a Glance

| Deadline | Authority | When | |---|---|---| | Last honest retention date for a full survey program | Practice, not rule | 6 months before opening expert disclosure | | Rule 26(f) conference — negotiate expert dates with survey lead time in mind | Fed. R. Civ. P. 26(f) | Per Rule 16(b) or local rule | | Opening expert disclosure and report | Fed. R. Civ. P. 26(a)(2)(D)(i) | 90 days before trial absent a stipulation or order | | Rebuttal expert disclosure | Fed. R. Civ. P. 26(a)(2)(D)(ii) | Within 30 days after the other party's disclosure | | Supplementation of expert disclosures | Fed. R. Civ. P. 26(e)(2) | When Rule 26(a)(3) pretrial disclosures are due — 30 days before trial | | Expert deposition time limit | Fed. R. Civ. P. 30(d)(1) | 7 hours, one day, absent stipulation or order | | Motion to exclude | Local rule / individual practices | Commonly with dispositive motions; check the meet-and-confer requirement | | Sanction for late or incomplete disclosure | Fed. R. Civ. P. 37(c)(1) | Exclusion unless substantially justified or harmless | | TTAB expert disclosure | 37 C.F.R. § 2.120(a)(2)(iii); TBMP § 401.03 | 30 days before the close of discovery | | TTAB pretrial disclosures | 37 C.F.R. § 2.121(e) | No later than 15 days before the testimony period opens | | TTAB testimony by declaration, subject to election of oral cross | 37 C.F.R. § 2.123(a) | Within the assigned testimony period | | Emergency survey for a TRO or PI | Fed. R. Civ. P. 65 | 4 weeks minimum, two cells, declaration not a Rule 26(a)(2)(B) report |