What changed to make damages exclusions more common?
Two things reinforcing each other. The federal rule governing expert testimony was amended at the end of 2023 to make explicit that the party offering an expert must establish admissibility by a preponderance, and that an opinion must reflect a reliable application of the method to the facts of the case. That clarified rather than raised the standard, correcting a drift toward treating these as questions of weight for the jury rather than admissibility for the judge.
Appellate courts have since applied that gatekeeping seriously in damages, which is where the inputs are frequently the softest part of an otherwise rigorous analysis. A damages model can be internally coherent, methodologically orthodox and still excluded, because the facts it was fed were assumed rather than established.
What actually happens if the damages expert is excluded?
Frequently something far worse than a reduced award. Where a patent owner has no admissible evidence supporting a royalty, courts have entered nominal awards even though infringement was established. In a 2025 appellate decision a jury award of roughly ten million dollars was reduced to one dollar, after the patent owner’s damages expert was excluded for relying on a license agreement without apportioning the value of the asserted patent from the other patents that agreement covered.
The structural point underneath that outcome is the one worth internalising: liability and damages are proved separately, and winning the first preserves nothing about the second. A case can be right on the merits and worth a dollar.
Is the trend only toward more exclusion?
No, and it is worth being accurate rather than alarmist about this. A 2026 appellate decision reversed a district court’s exclusion of a damages expert, holding that a usage-based royalty base — measured in virtual machine hours — did establish a sufficient causal connection to the accused functionality.
Read together, the direction is toward demanding a real evidentiary connection between the damages theory and the accused technology. That cuts against opinions that assume the connection and in favor of opinions that demonstrate it. It is a quality standard rather than a presumption against damages testimony, and it is available to defendants too — a rebuttal opinion resting on cherry-picked licenses is vulnerable on exactly the same grounds.
Can a case survive the exclusion of its damages expert?
Sometimes, and the outcomes are unforgiving enough that it is not a risk worth running. Where other admissible evidence in the record supports a royalty, a fact-finder may still award damages, and courts have occasionally permitted a further attempt rather than entering judgment.
Where the record contains nothing else, the result can be nominal despite established infringement. Because the answer depends so heavily on what independent evidence exists, the practical protection is structural rather than tactical: do not let the entire damages case rest on a single input that one ruling can remove.
When should the damages foundation be tested?
At the pleading and early discovery stage, which is far earlier than it usually happens. Almost everything that causes exclusion is knowable at the outset: what the license portfolio actually contains, whether component-level revenue exists, whether usage data is collected, whether the accused feature can be isolated at all.
The expensive version is discovering at expert disclosure that the evidence the opinion needed was never requested in discovery, by which point the schedule rarely permits a fix. Testing the foundation while it can still be built is the cheapest work in the case, and it is routinely deferred because it feels like a damages problem rather than a pleading problem.