What "Documented Witness Prep" Means to a Carrier After a Nuclear Verdict

By John Mahoney · August 2026 · 11 min read

There is a conversation that happens between a claims executive and a defense firm's relationship partner in the weeks after a catastrophic verdict, and it is not about the jury. Juries are unpredictable; every carrier prices that in. The conversation is about the file. Specifically: what does the file show that counsel did before the witness testified? If the answer is a time entry that reads "prepare witness for deposition — 4.0 hrs," the panel relationship is now in play, regardless of how good the actual preparation was. Work that is not documented, in the carrier's ledger, did not happen.

That question is being asked more often, at higher stakes, than at any point in the modern history of liability insurance. This piece walks through the numbers driving it, what carriers and their litigation management guidelines increasingly require, why the reptile playbook made witness performance the single most auditable variable in a defense file, and how documented preparation changes settlement posture — in both directions.

The Numbers Behind the Panic

The nuclear-verdict data is no longer a talking point; it is an underwriting input.

Marathon Strategies' Corporate Verdicts Go Thermonuclear counted 135 nuclear verdicts (≥$10M) against corporate defendants in 2024 — the highest on record — totaling $31.3 billion, delivered by juries in 34 states and 77 courts, with "thermonuclear" awards above $100 million nearly doubling year over year.
The trend did not pause: the firm's latest data, reported in August 2026, shows nuclear-verdict totals rising another 40.7% in 2025, with product liability alone accounting for 29 nuclear verdicts worth $12 billion and more industries hit than in any prior year of the study.
The U.S. Chamber Institute for Legal Reform's nuclear-verdicts research found that product liability, auto, and medical liability cases account for roughly two-thirds of nuclear verdicts, with a median nuclear verdict of $21 million and an average of $89 million over its 2013–2022 study window — and its May 2024 update documented both amounts and frequency continuing to climb.
In healthcare specifically, medical liability's share of high-dollar verdicts has grown to nearly 30% of reported nuclear verdicts, and The Doctors Company reports the average of the top 50 medical malpractice verdicts rose from $32 million in 2022 to $56 million in 2024 — a 75% increase in two years.

The paradox inside those numbers matters for defense strategy: claim frequency is flat to declining. Severity is doing all the work. Which means the marginal dollar of carrier loss is concentrated in a small number of cases that went badly at specific, identifiable moments — and post-verdict reviews keep finding the same moment: a defendant or corporate witness whose testimony armed the plaintiff's damages narrative.

Why Witness Performance Became the Audited Variable

The reptile approach — introduced in Ball and Keenan's 2009 manual and credited by its proponents with more than $8 billion in verdicts and settlements — is, at bottom, a witness-performance exploit. It does not attack the defense's experts or the medicine. It asks the defendant to agree to broad safety rules ("a doctor must never needlessly endanger a patient"), then walks the witness into agreeing that the rule was violated. We broke down the question pattern itself in The Reptile Strategy, Decoded; the defense-psychology literature — notably Kanasky's work for the IADC — is blunt that the countermeasure is not a clever objection at trial. It is witness training, before the deposition, against the specific failure modes that witness exhibits.

That is why the post-mortem after a nuclear verdict now runs through the witness file. Safety-rule agreement, over-explanation, adopting the examiner's characterizations, speculation outside the record — these are form failures, and form failures are countable. They can be counted in prior transcripts before prep begins, and they can be counted again afterward. Once something is countable, a claims organization will eventually demand that it be counted. That is the entire history of litigation management.

What Litigation Management Guidelines Increasingly Require

Anyone who has worked a panel file knows the guidelines document has been growing. Early-case assessments within 60 or 90 days. Budgets by phase. Pre-approval for experts, motions, and mock exercises. Adverse-development reporting. The direction of travel on witness preparation is the same, and firms are seeing versions of these requirements show up in guidelines, audit letters, and roundtable templates:

A named prep plan for each key witness — who prepares the witness, over how many sessions, covering what. "Witness will be prepared prior to deposition" is the kind of sentence bill review now flags rather than accepts.
Evidence that prior testimony was reviewed. If the defendant has testified before — and physicians, corporate representatives, and safety officers usually have — carriers want to see that someone actually read the old transcripts and identified what needs to change, not just that they were collected.
An assessment of the witness as a witness, separate from the merits: communication risk, volatility under cross, susceptibility to safety-rule framing. Several carriers now ask for this in the early-case assessment itself, because it drives settlement value as much as liability does.
Documentation proportionate to exposure. On a case reserved in seven figures, an undocumented prep process reads — after the fact, to an auditor — as a diligence gap, whatever the truth was.

None of this is exotic. It is the same evolution that happened to expert selection a decade ago: what used to be trusted judgment is now trusted judgment plus an artifact.

The Artifact: What a Prep Readout Actually Contains

A prep memo that satisfies both privilege discipline and carrier scrutiny has a recognizable shape. Built from the witness's prior transcripts and the prep sessions themselves, it documents:

Element	What it shows the carrier
Baseline: scored review of prior testimony (over-answering, adopted characterizations, speculation, safety-rule agreements — each cited to page and line)	Counsel identified the specific risks this witness carries, before prep began
Coaching plan mapped to each finding	Prep hours were structured against documented problems, not generic rehearsal
Post-prep or post-deposition delta — the same metrics, re-scored	The preparation measurably worked (or flagged residual risk that justified settlement)
Counsel-directed handling: retained by counsel, delivered to counsel, marked as work product	The artifact itself is defensible — see below

We covered the economics of producing this artifact — and why it fits inside capped, audited prep budgets when a human consultant does not — in The Deposition Readout Your Carrier Will Ask About. The short version: a scored readout is the difference between prep hours that survive bill review and prep hours that get written down.

A note on privilege discipline

Documentation only helps if it stays inside the privilege boundary. Prep materials generated at counsel's direction, for the purpose of rendering legal advice, are classic work product. Materials the witness generates alone — self-rehearsal notes, or sessions with consumer AI chatbots — may be neither privileged nor protected, a problem courts have already begun confronting with AI-assisted litigation materials prepared outside counsel's direction. The rule for the file is simple: every prep artifact should exist because counsel commissioned it, flow to counsel, and be labeled accordingly.

How Documentation Changes Settlement Posture

This is the part that gets missed when witness prep is treated as a compliance chore. A documented prep process changes the negotiation in three concrete ways:

1. It changes the carrier's willingness to try the case. Settlement decisions in the nuclear era are driven less by liability odds than by tail risk — the small probability of a catastrophic number. An unquantified witness is pure tail risk. A witness with a scored baseline, targeted coaching, and a measured improvement is a bounded risk, and bounded risks get tried. Conversely, a readout that shows residual, untrainable volatility is exactly the evidence a claims committee needs to authorize settling early, before the deposition creates the transcript that plaintiff's counsel will read to the jury. Both outcomes are wins; both require the measurement.

2. It changes what the plaintiff can do with the deposition. Mediators price cases off deposition excerpts. The safety-rule concessions and over-explanations that the reptile sequence is designed to harvest are precisely the excerpts that move mediation brackets by seven figures. Prep that demonstrably closes those doors does not just improve trial odds — it removes the raw material of the plaintiff's settlement leverage.

3. It protects the firm when the outcome is bad anyway. Verdicts will still go wrong — 135 times in 2024, by Marathon's count. When one does, the difference between "we prepared the witness" and a dated, scored, work-product readout showing exactly what was found and fixed is the difference between a post-mortem and a panel review. Defense firms do not lose panels over verdicts; they lose them over files that suggest the verdict was foreseeable and nothing in the record shows anyone looking.

What This Looks Like in Practice

For a panel firm or a self-insured health system's staff counsel office, the minimum credible workflow on any case with meaningful exposure:

Pull every prior transcript the witness has. Depositions in other matters, trial testimony, peer-review or administrative proceedings where obtainable.
Score the baseline before the first prep session. Count the failure modes: unresponsive elaboration, adopted characterizations, speculation, absolute statements, safety-rule agreements. Cite each to page and line.
Run prep against the findings, not against a generic checklist — and record which findings each session addressed.
Re-score after the deposition and put the delta in the file, alongside the reporting the guidelines already require.
Route everything through counsel and mark it as work product, with retention handled accordingly.

Done manually, step 2 is hours of associate time per transcript — which is why it historically didn't happen, and why "prepared witness, 4.0 hrs" became the industry's default documentation. Done with structured analysis, it is an afternoon and a fixed fee, which is the point at which the documentation standard stops being aspirational.

The readout carriers are starting to ask for — from one transcript

A scored witness-performance audit: every over-answer, adopted characterization, and safety-rule concession cited to page and line, with a coaching plan. Counsel-directed, work-product handling, 48-hour turnaround.

See the Witness Performance Audit →

The Uncomfortable Summary

The nuclear-verdict era did not change what good witness preparation is. Defense lawyers have always known that the case often dies in the defendant's deposition. What changed is who is looking, and what they expect to see. Carriers absorbed $31 billion in nuclear verdicts in a single year and responded the way risk institutions always respond: by demanding that the highest-variance input in the file — the witness — be measured, managed, and documented like every other one.

Firms that treat that as bureaucratic overhead will keep writing "prepare witness — 4.0 hrs" and hoping. Firms that treat it as what it is — the cheapest available protection for both the case outcome and the panel relationship — will have the artifact ready before anyone asks. After the next verdict in your jurisdiction goes nuclear, only one of those firms is having a comfortable conversation with the claims executive.