Lex Machina

What it is

Lex Machina is a litigation analytics platform: it turns court dockets and filings into structured data about how judges rule, how long cases take, what damages get awarded, and which firms win for which clients. It began as an IP litigation project at Stanford, spun out in 2010, and has been owned by LexisNexis since November 2015. LexisNexis still sells it as its own product, with its own login and sales process, not as a feature of Lexis+.

The data is the product. LexisNexis says Lex Machina covers all 94 federal district courts, all 13 federal courts of appeal, PTAB, specialty venues and a set of “enhanced” state courts, built from 45 million documents across more than 10 million cases, 8,000+ judges and 6,000+ expert witnesses (vendor figures as of April 2025). Attorney reviewers code outcomes, damages and findings on top of the raw docket, which is why its numbers are more defensible than a keyword count of docket entries.

Since April 2025 Protégé sits on top of it. LexisNexis added its Protégé assistant to Lex Machina on 8 April 2025, so a user can type “how often does this judge grant summary judgment in contract cases?” and get the analytic back without building filters. The rest of the past year added depth: docket-level data for 1,200 more state trial courts drawn from CourtLink (November 2025), an analytics API positioned for internal dashboards and client portals (March 2026), and enhancements to Litigation Footprint, the view that combines a party’s state, federal and administrative litigation (September 2026).

Why it shows up in Legal Ops stacks

• Benchmarking outside counsel with outcomes, not pitch decks. In-house teams use counsel and law-firm analytics to see which firms actually win motions, trials and appeals in a practice area and in front of a given judge. The 2026 Law Firms Activity Report (16 June 2026) is built for exactly that: it found ten private firms each won more than $1 billion in principal damages through plaintiffs’ jury verdicts in federal court between 2023 and 2025.
• Early case assessment and litigation budgets. Timing events, damages by case type and class action settlement data (attorney fees, class representative awards) give a reserve or budget a number to anchor on before the first status conference.
• Patent and IP work, where it started. Federal patent coverage plus PTAB is still its deepest vertical. Its 2026 Patent Litigation Report (5 August 2026) analysed 38,893 district court patent cases filed from 2016 through 2025, and LexisNexis counts 4,547 new district court patent cases in 2025, nearly 20% above 2024.
• Feeding your own dashboards. The API is the route for a legal ops team that wants judge and counsel metrics inside a matter-management or BI tool rather than in another research tab.

Pricing reality

There is no published price and no self-serve trial — only a guided demo and a quote. Lex Machina has historically been sold by practice-area and state modules; its 2021 Enterprise Package bundled all modules with unlimited users inside an organisation. Quotes scale with the courts and practice areas you license, seats, and API access, and it is not listed in LexisNexis’s small-firm online store.

Third-party estimates put small-firm and small-department contracts in the tens of thousands of dollars a year and AmLaw-scale enterprise deals in six figures. Treat those as unverified; no vendor or procurement source confirms them. The practical rule: if a quote is priced like a per-attorney add-on to your Lexis+ subscription, check whether you are being offered Lex Machina or the lighter Litigation Analytics included in some Lexis+ tiers, which is a different product.

Best for

In-house litigation counsel and legal ops leads at companies with recurring federal litigation who select, benchmark and negotiate with outside firms, and litigation partners and practice-support teams at mid-size and large firms who need defensible judge, counsel and damages data for venue strategy, pitches and budgets.

Not for: practices that live in state trial courts outside the enhanced set, transactional teams, or anyone whose budget is a few hundred dollars a month. It supplies analytics, not case-law research — you still need a primary research platform.

Alternatives and when to pick them

• Westlaw Litigation Analytics (Thomson Reuters, sold with Westlaw Advantage research subscriptions) — the market-share incumbent. Pick it when the department already pays Thomson Reuters and the analytics questions are occasional; the analytics ride on a contract you already have.
• Bloomberg Law — the other installed-base option, with litigation analytics alongside dockets and news. Pick it when docket tracking at volume and Bloomberg’s business data matter as much as judge analytics.
• Trellis — the fastest-growing entrant, built around state trial courts in 45 states, with self-serve pricing from $69.95 a month and MCP connectors for Claude and ChatGPT. Pick it when your matters sit in county courts or the budget is per-attorney, not enterprise.
• CourtListener — free federal dockets and opinions with an MCP server. Pick it when you need raw docket data for a Claude workflow and can do the counting yourself.
• Darrow — pick it for finding cases that have not been filed yet; Lex Machina explains how filed cases behave.

Watch-outs

• State coverage is two-tier. “Enhanced” state courts carry coded motion metrics for 40+ motion types; the 1,200 courts added from CourtLink carry docket-level data only. Guard: send the sales team your top ten venues and get, in writing, which are enhanced and which are docket-only before you sign.
• Lex Machina and Lexis+ Litigation Analytics are easy to confuse. They are separate products with different data depth. Guard: the order form should name Lex Machina explicitly and list the licensed modules, courts and seat count.
• Generative answers still need checking. Protégé returns analytics from structured data, but a 2024 Stanford RegLab study measured hallucination rates above 17% on Lexis+ AI, Protégé’s predecessor. Guard: for any figure going into a client memo, budget or panel decision, open the underlying case list and cite the cases, not the chat answer.
• Statistics hide small samples. A judge’s 80% grant rate on twelve motions is not a trend. Guard: set a minimum sample size (we use 30 rulings) before a metric goes into a recommendation, and show the n alongside the percentage.

For the wider category, see best legal AI research tools and the LexisNexis Protégé page.