Opposing counsel opened the 80-page expert report on a Friday afternoon. By Monday morning, they had filed a Daubert motion. The target was not the expert’s conclusion. It was a single sentence on page 23 where the expert described a methodology as “widely accepted” without a single source establishing that acceptance.
The expert had spent three months on the report. Opposing counsel spent two hours finding the crack.
That is the asymmetry of expert witness work. The report is built over months. It is attacked in hours. And the attack does not target the opinion, it targets the scaffolding underneath.
The Six Axes Opposing Counsel Targets
They do not read the report the way a judge reads it. They read it the way a structural engineer reads a bridge, looking for the load-bearing weakness.
Axis 1: Methodological Consistency
Is the method applied consistently throughout the report, or does the expert use different standards for favorable and unfavorable evidence?
Opposing counsel checks:
- Does the expert apply the same methodological rigor to all data, or only to data that supports the opinion?
- Are the methods described in the methodology section actually the methods used in the analysis?
- Does the expert cite methodological standards and then deviate from them without explanation?
The vulnerability: an expert who describes Method A in the methodology section but applies Method B in the analysis is giving opposing counsel a clean Daubert argument, the methodology as described was not the methodology as applied.
Axis 2: Citation Accuracy
Does every cited source actually support the claim it is attached to?
This is where the 156 documented lawyer sanctions for AI hallucinations and the Kohls v. Ellison exclusion precedent matter most. Opposing counsel checks citations not for formatting, for alignment.
They verify:
- Does the cited paper exist? (DOI resolution, the easiest check)
- Does the paper support the claim, or is it a backwards citation, real paper, inverted finding?
- Is the paper contested in the literature? A 35% contradiction ratio means the paper is not settled evidence.
- Is the paper being applied to a claim it does not make? Structural misuse, a methods paper cited as a clinical outcome.
The expert who used AI assistance in preparing the report and did not verify the citations is in the same position as the expert in Kohls v. Ellison. The exclusion was not about the expert’s conclusion. It was about the verifiability of the sources.
The expert report that passes internal review may not pass The Gate, the verification threshold that catches methodological vulnerabilities before opposing counsel applies pressure.
Axis 3: Logical Completeness
Does the analysis account for alternative explanations, or does it present a single-path argument?
Opposing counsel looks for:
- Did the expert test alternative hypotheses and rule them out, or only test the favored hypothesis?
- Are there obvious confounding variables that the analysis does not address?
- Does the report acknowledge limitations of the data or methodology, or does it present the analysis as definitive?
The vulnerability: an expert who only tests the favored hypothesis is not doing science. They are doing advocacy. And advocacy dressed as methodology is what Daubert motions are built on.
Axis 4: Assumption Transparency
Are the assumptions underlying the analysis stated clearly, or buried in appendix footnotes?
Opposing counsel checks:
- What assumptions does the model require? Are they stated? Are they reasonable?
- If an assumption is wrong, does the conclusion collapse? If yes, was the assumption tested?
- Does the expert acknowledge the sensitivity of the conclusion to its underlying assumptions?
The vulnerability: an expert whose conclusion depends on an unstated or untested assumption is building on a foundation opposing counsel can excavate.
Axis 5: Data Provenance
Where did the data come from? Is its chain of custody documented? Is it the right data for the question?
Opposing counsel checks:
- Was the data collected using standard methods? If not, why?
- Is there a documented chain of custody from source to analysis?
- Is the data appropriate for the question the expert is answering? A dataset designed for one population used to make claims about another is a provenance vulnerability.
This is the axis that takes the most time to investigate and produces the most damage when it fails. If the data is flawed, the analysis built on it is compromised regardless of how well the analysis was executed.
Axis 6: Conclusion Support
Does the data actually support the strength of the conclusion, or has the expert overstated certainty?
Opposing counsel checks:
- Does the data show correlation where the expert claims causation?
- Are the confidence intervals wide enough that the conclusion is less certain than the language suggests?
- Does the expert’s language (“the data demonstrate”) exceed what the design supports (“the data suggest”)?
The vulnerability: expert overreach. The expert who has formed a strong opinion is tempted to let the language reflect certainty the data does not carry. “Suggest” becomes “demonstrate.” “Associated with” becomes “causes.” That shift is not rhetorical, it is methodological, and it is what exclusion motions are built on.
The Contradiction Register
The vulnerability scan that runs before opposing counsel runs their version.
The Contradiction Register receives a completed expert report and returns a structured assessment across all six axes. Each axis produces flags with severity ratings, specific document locations, and recommendations:
CITATION ACCURACY, FLAGGED
Citation [23]: Johnson et al. (2019) cited for dose-response.
Source found no significant effect (p=0.18). Backwards citation.
Severity: HIGH. Replace or reframe.
CONCLUSION SUPPORT, YELLOW FLAG
Page 67: "data demonstrate causation" exceeds correlational design.
Severity: MEDIUM. Reframe to "demonstrate association."
METHODOLOGICAL CONSISTENCY, FLAGGED
Page 23: Method described as "widely accepted", no source.
Severity: HIGH. Cite peer-reviewed consensus or reframe.This is not a grade. It is a register, every flagged item, every severity, every recommendation. The expert and their retaining counsel decide which flags to address and how. The register does not rewrite the report. It tells the people who wrote it where opposing counsel will look.
The Cost of Not Running the Register First
The expert who finds the vulnerability before opposing counsel costs $0 in additional fees. The expert who finds it during a Daubert hearing costs $50,000+ in replacement fees, lost settlement positioning, and weakened credibility.
The report in the opening scenario was excluded. The expert’s $50,000 in fees was unrecoverable. The case settled on weaker terms. The expert’s credibility in that jurisdiction was damaged.
All of it traced back to one sentence on page 23 that no one stress-tested before filing.
Opposing counsel does not read your report to understand your opinion. They read it to find the weakest link. The expert who simulates that read before the report is filed is the expert whose report survives the Daubert hearing.
If your next expert report has not been stress-tested against the six axes opposing counsel targets, the courtroom will do it for you. Request a legal intake.