Product-Liability Litigation: Experts and the Document Record

By John Mahoney · June 2026 · 9 min read

A product-liability case is, at bottom, a contest over a corporate paper trail. Whether the theory is a defective design, a flawed unit off the line, or an inadequate warning, the question the jury has to answer — was this product unreasonably dangerous, and did the defect cause the harm — is almost always decided by what the manufacturer's own records show and what its experts can say about them. The engineer or metallurgist who reasons cleanly from the design history file, the test data, and the complaint record is formidable. The one who offers an alternative design they never built or tested, or a failure theory the documents do not support, is a Daubert motion waiting to happen.

This guide walks through the three defect theories and the legal tests that govern them, the document and evidence record that decides most cases, the expert disciplines involved, how the methodology gets attacked, and the signals that separate a strong opinion from a weak one — for plaintiff and defense attorneys evaluating a product file.

Disclaimer: This article is for informational purposes only and does not constitute legal advice. Product-liability theories, the applicable defect test, and admissibility standards vary by product, jurisdiction, and facts, and the controlling reliability standard is jurisdiction-specific. Always verify the controlling standard and the underlying record before relying on any generalization.

The Three Defect Theories — and the Legal Tests

Product-liability claims sort into three classic theories, and the theory determines both what the expert must prove and which body of documents matters most.

• Design defect. The product was built as intended, but the design itself is unreasonably dangerous — every unit shares the flaw. This is the theory that turns on engineering judgment and, in most jurisdictions, on a feasible alternative design.
• Manufacturing defect. The design is sound, but this particular unit departed from it — a bad weld, contaminated material, an out-of-spec part, a missed inspection. The benchmark is the manufacturer's own specification, so the case lives in the QA and production records.
• Failure to warn (marketing defect). The product carries a non-obvious risk that adequate instructions or warnings could have mitigated, and the warnings given were absent, inadequate, or poorly placed. This theory turns on what the manufacturer knew, when, and what it told the user.

Two legal tests recur in design cases, and which one controls is jurisdiction-specific. The risk-utility test asks whether the design's risks outweigh its benefits — weighing the gravity and likelihood of harm against the burden of a safer alternative, the product's utility, and cost. It typically requires the plaintiff to show a reasonable alternative design that was feasible and would have reduced the risk without destroying the product's usefulness. The consumer-expectation test asks instead whether the product was more dangerous than an ordinary consumer would expect. Some jurisdictions use one, some the other, and some apply both or pick based on the product's complexity. The choice matters enormously: a risk-utility jurisdiction puts the alternative-design proof at the center of the case, while a consumer-expectation framing can shift the fight to ordinary user assumptions.

The Document and Evidence Record

More than in almost any other case type, the product-liability record is the case. The defect, the manufacturer's knowledge, and frequently the alternative design are all proved — or disproved — out of the company's own files. The categories that decide most matters:

• The design history file. The engineering record of how the product came to be — design inputs and outputs, specifications, drawings, design reviews, change orders, and design-verification and validation testing. This is where a feasible alternative either appears (the company considered and rejected a safer design) or does not, and it is the first thing both sides mine.
• Testing and QA records. Design-validation tests, durability and stress testing, inspection records, statistical process controls, lot and batch records, and acceptance criteria. In a manufacturing-defect case these define the spec the unit was supposed to meet; in a design case they show what the company tested and what it learned.
• Complaint and recall history. Prior incident reports, warranty claims, customer complaints, field-failure data, prior lawsuits, and any recall, retrofit, or service bulletin — together with regulatory submissions and adverse-event reporting where applicable. This record speaks directly to notice: when did the manufacturer learn of the risk, and what did it do.
• The exemplar and the failed product. The actual unit at issue and an exemplar (an identical, undamaged unit) for comparison and testing. Preservation and chain of custody are decisive here — destructive testing of the failed unit has to be coordinated, and a unit that is altered, repaired, or lost becomes a spoliation problem.
• Internal communications. Engineering emails, meeting minutes, risk assessments, and cost-benefit memoranda. This is where intent and knowledge live — a document showing the company weighed a known hazard against the cost of fixing it is often the most consequential page in the file.
• Warnings, labels, and instructions. The actual on-product warnings, manuals, packaging, and marketing materials, plus any internal review of warning adequacy and the applicable labeling standards. In a failure-to-warn case this is the core exhibit set.

The practical lesson for the attorney: the case is decided early, by what gets preserved and what gets produced. A preservation letter that locks down the failed unit and an exemplar, and a document-request strategy that reaches the design history file, the complaint history, and the internal risk and cost memoranda, is often worth more than any single expert retention later.

The Expert Disciplines

Product cases are multi-expert by nature, and matching the discipline to the theory is half the battle.

• Mechanical / design engineering. The central design-defect witness — opines on whether the design was unreasonably dangerous and, critically, whether a reasonable alternative design was feasible. Credibility rises sharply when the alternative was actually built and tested rather than merely sketched.
• Metallurgy and failure analysis. The witness who determines why the part failed — fatigue, overload, corrosion, a material or weld defect — through fractography, metallurgical examination, and (often destructive) testing of the failed component. This discipline frequently distinguishes a manufacturing defect from a design flaw from ordinary wear or misuse.
• Human factors and warnings. The witness on foreseeable use and misuse, warning adequacy and placement, and whether a different warning would have changed user behavior. Essential to failure-to-warn theories and to defending against them.
• Industry-standard and regulatory experts. Witnesses who establish the applicable consensus standards, codes, and regulatory requirements, and whether the product met them — relevant to both the standard of care and the state of the art at the time of design.

The recurring qualification trap is scope. A design engineer is not automatically a metallurgist, and neither is a human-factors expert. An expert who strays outside the lane their background supports is the easiest one to limit or exclude. Match each expert's demonstrated expertise to the specific opinion the case actually needs, and be wary of a single witness asked to carry all three theories.

How the Methodology Gets Attacked Under Daubert

Product-liability opinions are routinely the subject of reliability challenges, and the attack is almost always aimed at the gap between the record and the conclusion rather than at engineering itself. The recurring lines of attack:

• The untested alternative design. The most common and most effective attack in design cases — the expert proposes a safer alternative but never built, tested, or validated it, and cannot show it would have prevented this failure without introducing new risks or destroying the product's utility.
• Failure analysis without testing. A failure-mode opinion offered without the metallurgical or destructive testing that would actually distinguish fatigue from overload from corrosion — conclusion first, examination never.
• Ignoring the spec or the data. A manufacturing-defect opinion that does not measure the unit against the actual specification, or a design opinion that discounts the company's own validation testing without a principled reason.
• Cherry-picking the record. Crediting the complaint or test data that supports the retaining party while discounting contrary entries — the prior incident that does not fit, the test that passed — without explanation.
• Ipse dixit. The conclusion does not follow from the stated method; the analytical gap between the documents and the opinion is too great, a classic basis for exclusion.
• Reaching beyond the discipline. Design qualifications stretched into metallurgy, or engineering credentials stretched into human-factors or warning-adequacy opinions the expert is not qualified to give.

The reliability inquiry is methodology-focused, and the strongest defense against exclusion is an opinion that tests its alternative design, grounds its failure analysis in actual examination, measures the unit against the real specification, and ties every step to the produced record. The strongest attack is showing the opinion rests on an alternative no one ever built or a failure mode no one ever tested.

Strong Signals vs. Weak Signals — for Both Sides

Both sides screen the same record for the same tells, and the factors separating a strong product opinion from a vulnerable one are largely symmetrical.

What strengthens a product opinion

• A reasonable alternative design that was actually built and tested, shown to reduce the risk without destroying utility.
• A failure analysis grounded in metallurgical or destructive testing that distinguishes the failure mode from wear, misuse, and ordinary use.
• A manufacturing-defect theory measured against the manufacturer's own specification, lot records, and acceptance criteria.
• A complaint, recall, and internal-communication record that establishes notice and is read in full, not selectively.
• An expert opining squarely within their demonstrated discipline, with the failed unit and an exemplar preserved.

What weakens a product opinion

• An alternative design that exists only on paper — never built, never tested, never validated against new risks.
• A failure theory asserted without the metallurgical examination or testing that would actually support it.
• A defect claim that never compares the unit to the spec, or a design opinion that ignores the company's own validation data.
• A reading of the complaint or test record that credits the favorable entries and discounts the unfavorable ones without a reason.
• A failed unit altered, repaired, or lost before inspection (a preservation and spoliation problem), or an expert reaching beyond their discipline.

The defect opinion often decides the case — so the disciplined work for either side is to separate the links carried by the produced documents and actual testing from the links carried by assumption, and to attack or defend the opinion at exactly those joints.

Why a Verifiable, Page-Cited Review of the Underlying Records Matters

A product case is built on a mountain of documents — the design history file, the testing and QA records, years of complaint and warranty data, recall and service bulletins, internal engineering emails and risk memoranda, the warnings and manuals, and the expert reports and depositions on both sides. The opinion is only as good as those underlying records, and the case is won or lost on whether each expert claim is actually grounded in them. That makes the document-review step decisive: which test result appears on which page of the validation report, what the design-review minutes actually recorded, where the opposing expert's report quietly assumes an alternative the file shows the company rejected for a reason, and whether a deposition admission contradicts the written opinion.

This is exactly where a verifiable, page-cited AI review earns its place. MedLegal AI reviews the full record and ties every finding back to the exact source page — so when the system flags a complaint the company received before the incident, an internal memo weighing a known hazard against cost, or an assumption the testing data undercuts, you can click straight to the page and confirm it. Nothing is asserted without a citation, which is the difference between a usable cross-examination outline and an AI summary you cannot trust. The platform now supports non-medical cases — including product-liability files — and connects litigators to a network of non-medical expert witnesses.

Bottom Line

Product-liability cases turn on the corporate document record — the design history file, the testing and QA data, the complaint and recall history, the internal communications, and the warnings — and on whether the experts can ground their opinions in it. The three defect theories (design, manufacturing, failure to warn) each demand a different proof and a different body of documents, and which legal test controls a design claim (risk-utility, often with a reasonable alternative design, versus consumer-expectation) shapes the whole case. Credentials matter — design engineering, metallurgy and failure analysis, human factors, and industry standards among them — but scope matters more, and the Daubert attack almost always targets the gap between the produced record and the conclusion, above all the alternative design no one built and the failure mode no one tested.

For both sides, the work is the same: confirm what was preserved and produced, separate the links carried by documents and testing from the links carried by assumption, hold each expert inside their demonstrated discipline, and read the complaint, test, and internal-communication record in full rather than selectively. The record should decide the case — so verify every claim against the actual file, page by page.

Questions? Contact us at [email protected] or (856) 979-6525