LexHow we rank patent litigation.
Six performance signals drawn from PACER district court dockets. Draft methodology — the data pipeline is built, peer review is ongoing.
Draft · Coming 2026Last updated May 12, 2026
Patent litigation rankings have to disaggregate plaintiff and defendant work. A firm that wins 70% of its patent-owner cases is doing different work from one that wins 70% of accused-infringer cases. The methodology publishes both streams separately.
Every signal traces back to PACER docket data, normalized against the Patexia Patent Litigation index that we have maintained since 2014. Markman orders, summary judgment dispositions, trial verdicts, and post-trial motions are all parsed and classified.
This is a draft methodology. The signal definitions and weights below are stable enough for internal use but will be finalized after a 90-day peer review window. Email rankings@lexdana.ai with feedback.
Six signals. Six weights.
Outcomes-led, with Markman and verdict signals carrying the most weight.
Case volume
Total cases as lead or co-counsel within the window. Entry threshold and tiebreaker.
Markman outcomes
Claim construction wins (favorable construction obtained, key terms preserved). Normalized for judge tendencies.
Settlement vs. trial rate
Settlement timing and terms relative to case posture. Settling well is a skill; it’s counted separately from winning at trial.
Trial verdicts
Jury and bench verdicts, weighted by the size of the case (asserted damages and number of patents in suit). The dominant signal.
Damages outcomes
Damages awarded vs. damages sought, plus willfulness and enhanced-damages findings. Captures recovery-quality, not just win-rate.
Time-to-resolution
Filing-to-judgment days, normalized for case complexity and number of asserted patents.
How the math works.
Every signal is normalized within a peer cohort, then weighted into a composite score.
Cohort-relative percentiles, not absolute counts
Within each ranking, firms and attorneys are grouped into two streams (plaintiff practice, defendant practice) plus four firm-size brackets. Inside each cohort, every signal is converted to a percentile rank.
This is why a 30-attorney specialty boutique can outrank a global AmLaw firm on the same signal — volume alone doesn’t win the comparison.
Weighted blend, capped at 100
Each signal’s percentile is multiplied by its weight, then summed. Final scores are rescaled to a 0–100 range so cross-cohort comparison is meaningful.
Worked example: (0.76 × 25) + (0.69 × 20) + …
What’s included.
- Firms with at least 25 patent cases in the 5-year window
- All US district courts; both jury and bench trials
- Hatch-Waxman (ANDA) cases excluded — see ANDA methodology
- Lead, co-lead, and significant trial-team counsel
- Multi-district litigation (MDL) treated as one case per defendant
What we don’t measure.
- ITC investigations (separate ITC methodology)
- PTAB-only practice (separate PTAB methodology)
- International forum cases (UPC, China, Germany, etc.)
- Pure pre-suit licensing work without filed cases
- Reputation, peer votes, or pitch quality
What changed — and when.
Methodology changes are dated and archived. The Draft v0.3 edition remains accessible exactly as published.