---
title: "Legal Research Platforms and Tools: Westlaw, Lexis and Free Options"
seoTitle: "Legal Research Platforms: Westlaw, Lexis & Free Tools"
url: https://www.masstortmarketingagency.com/blogs/legal-research-tools
canonical: https://www.masstortmarketingagency.com/blogs/legal-research-tools
published: 2026-10-22
modified: 2026-10-22
author:
  name: Tarun
  role: Founder, Mass Tort Marketing Agency
publisher:
  name: Mass Tort Marketing Agency
  url: https://www.masstortmarketingagency.com
description: |
  Legal research platforms compared on price, coverage and citators: Westlaw, Lexis, Bloomberg, free sites and state bar access, plus a stack for PI firms.
keywords:
  - legal research platforms
  - legal research tools
  - online legal research platform
  - legal research software
  - Westlaw vs LexisNexis
  - free legal research sites
  - Lexis+ with Protégé
  - vLex Fastcase
  - AI legal research tools
banner: /images/blog-content/legal-research-tools/hero.jpg
ogImage: /images/og/legal-research-tools-og.jpg
license: |
  Cite freely with attribution to Mass Tort Marketing Agency. Verbatim quoting
  permitted with citation back to the canonical URL.
---

> **Quick answer.** The leading legal research platforms are Westlaw, Lexis+ with Protégé and Bloomberg Law. vLex Fastcase (now part of Clio) and Decisis are the low-cost services that every state bar gives members free, per [Duke Law's May 2025 state bar table](https://law.duke.edu/lib/statebarassociations). Free sources such as Google Scholar, CourtListener, Cornell LII, govinfo and PACER's free opinions cover a lot of ground but lack a full citator. Westlaw's published plans start at $155.35 a month for small firms, as of October 2026. Check every AI answer with a citator.

**Key takeaways**

- Westlaw sells [three versions](https://legal.thomsonreuters.com/en/westlaw) (Edge, Edge with AI-Assisted Research, Advantage). [Published prices start at $155.35 a month](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge-aar/plans-pricing?segment=lawfirm) on a 3-year term and [apply only to new customers at firms with up to 10 attorneys](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge/plans-pricing?segment=lawfirm), as of October 2026.
- [Lexis+ AI is now Lexis+ with Protégé](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page) (February 2026). Pricing is customized, [1 to 3 attorney firms can buy online](https://www.lexisnexis.com/en-us/small-law-firms.page), and there is a 2-day trial.
- Casetext is no longer a standalone product, and Fastcase, vLex and Docket Alarm have been part of Clio since [November 10, 2025](https://www.clio.com/about/press/clio-completes-landmark-1b-vlex-acquisition-series-g-5b-valuation/).
- Every state bar offers free vLex Fastcase or Decisis, per Duke Law's table; 10 states are Decisis-only.
- [PACER charges $0.10 a page](https://pacer.uscourts.gov/pacer-pricing-how-fees-work) with a $3 cap per document, waives quarters of $30 or less, and never charges for opinions.
- A [peer-reviewed study](https://arxiv.org/abs/2405.20362) found leading AI research tools hallucinated 17% to 33% of the time in 2024 versions, and [ABA Formal Opinion 512](https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf) ties verification to competence.
- Plaintiff firms need more than case law: the JPML listed [162 pending MDL dockets on September 30, 2026](https://www.jpml.uscourts.gov/sites/jpml/files/Pending_MDL_Dockets_By_Circuit-September-30-2026.pdf), and FDA, CPSC and NLM databases feed product cases.

The legal research market has changed more since 2023 than in the decade before it. Products were bought, merged, renamed or retired, and the market map below lists each change with its date. Most "best legal research tools" lists have not caught up: many still recommend a product you can no longer buy or quote a price that no longer exists. The right answer for your firm depends on three things this page keeps returning to: firm size, practice area and budget.

This guide is written for solo and small-firm lawyers, personal injury and mass tort firms, litigation associates and paralegals, and law students who need the free sources. It runs in this order: what a platform is and the quick picks, the 2026 market map, a research workflow, the paid platforms in depth, free and bar-benefit sources, docket, analytics and verdict tools, AI research, the plaintiff-firm research stack, how to choose, firm operations, and an FAQ. Every price and feature was checked on the vendor's own page on October 22, 2026. Nothing here was lab-tested or benchmarked.

A disclosure: Mass Tort Marketing Agency is a [personal injury lawyer marketing](/personal-injury-lawyer-marketing) and mass tort lead-generation agency, not a research vendor, and it has no affiliate or referral arrangement with any vendor named here. The plaintiff-side chapter covers the research that sits behind deciding which torts to pursue and which claimants qualify.

## What a legal research platform is and does

A legal research platform is software that lets lawyers find, verify and apply primary law (cases, statutes, regulations and court rules) and secondary commentary. It combines search, a citator that flags bad law and, increasingly, dockets, analytics and AI assistants. LexisNexis describes the category as tools that ["assist legal professionals in finding and analyzing legal information, such as case law and statutes, to support legal decision-making."](https://www.lexisnexis.com/en-us/solutions/legal-research.page)

That definition covers a wide range of products. At one end sit full-service subscriptions with editorial headnotes, treatises, practice guides, a citator and an AI assistant. At the other end sit free government and nonprofit sites that publish the text of opinions and statutes with little or no editorial layer. Between them are bar-benefit services, specialist docket and analytics tools, and low-cost AI research products. The sections below sort them into terms, categories and authority types so the rest of the guide reads cleanly.

### Platform, tool, database or software: the terms

Vendors and reviewers use these words loosely, and the difference matters when you compare quotes.

- A **database** is a body of content, such as a case-law collection, a state's annotated code or a docket archive.
- A **tool** is a single function: a citator, a docket alert, a verdict search or a clause library.
- A **platform** is a subscription that bundles content, search, a citator and add-on tools under one login and one contract.
- **Legal research software** is the buyer's umbrella term for all of the above.

The practical point: buyers compare platforms by features, but they pay for content scope. A plan that covers one state and one federal circuit is a different product from one that covers all states, federal courts and a secondary-source library, even if the interface is identical. Read the coverage list before the feature list.

### The seven categories of legal research tools

Most firms do not need every kind of research tool, but every firm needs to know which kind answers which question. The table sorts the market into seven categories by the question each answers. Examples are names only; Parts 2 and 3 describe what each vendor says its product does.

| Category | Question it answers | Examples | Typical price model (Oct 2026) |
|---|---|---|---|
| Primary law research | What is the law? | Westlaw, Lexis+, Fastcase Library, Google Scholar | Quote, published plans, bar benefit or free |
| Citators | Is it still good law? | KeyCite, Shepard's, Cert Citator, Cytator | Bundled with the platform |
| Secondary sources and practice guidance | What does the law mean in practice? | Practical Law, Practical Guidance, treatises | Bundled or add-on |
| Dockets and court records | What is happening in a case? | PACER, CourtListener with RECAP, CourtLink, Docket Alarm | Per page, free or quote |
| Litigation analytics | How do this judge, court or counsel behave? | Lex Machina, Westlaw and Lexis analytics | Quote |
| Verdict, settlement and expert research | What is this case worth, and who is the expert? | Lexis Verdict & Settlement Analyzer, Lexis Expert Research On-Demand | Quote |
| Investigative and contract research | Who and where is this person, or what does this clause say? | Tracers, CLEAR, Law Insider | Quote, or $1,790 per user per year |

The citator names come from the vendors: Thomson Reuters' [KeyCite](https://legal.thomsonreuters.com/en/products/westlaw/keycite), LexisNexis's Shepard's, the [Cert Citator in Fastcase Library](https://vlex.com/fastcase-library) and [Descrybe's Cytator](https://descrybe.com/). LexisNexis lists [CourtLink, its Verdict & Settlement Analyzer and Expert Research On-Demand](https://www.lexisnexis.com/en-us/solutions/legal-research.page) as parts of its research stack, and Law Insider's SimpleAI plan is [$1,790 per user per year after a 7-day trial](https://www.lawinsider.com/pricing), as of October 2026; confirm with the vendor.

Most firms buy one platform that covers the first three categories and add the others as their practice demands. Parts 2 to 4 cover each layer in turn. AI research assistants are not listed as an eighth category because they sit on top of categories one to three: they search primary law, summarize secondary sources and, at best, call the citator. Part 4 covers how they work and how to check them.

### Primary and secondary, mandatory and persuasive authority

Two distinctions decide what you can cite and how much weight it carries. Cornell's Wex defines [primary authority](https://www.law.cornell.edu/wex/primary_authority) as the law itself: constitutions, statutes, regulations and court decisions. [Secondary authority](https://www.law.cornell.edu/wex/secondary_authority) explains or comments on the law, and Wex says it is "never mandatory." The second distinction is weight. [Mandatory authority](https://www.law.cornell.edu/wex/mandatory_authority) is "binding and must be followed," and whether a source is mandatory depends on the court and jurisdiction. [Persuasive authority](https://www.law.cornell.edu/wex/persuasive_authority) can influence a court without binding it, and Wex lists "Restatements, treatises, and jury instructions" among its forms.

A worked example. A plaintiff firm in one state is building a design-defect claim and finds a well-reasoned appellate decision from a neighboring state that adopts the same defect theory. That decision is primary authority, because it is a court decision. At home, it is only persuasive, because the home state's courts are not bound by a sister state. If the home state's own supreme court has spoken on the theory, that decision is mandatory and controls, even if the sister-state opinion is better written.

The distinction drives tool choice. Secondary sources are the fastest way into an unfamiliar issue, because they summarize the law and point to the leading cases. Primary law is what you cite in a brief. The citator tells you whether that primary law still holds. A research stack that skips any of the three leaves a gap: no orientation, nothing citable, or no way to know that a case was reversed last year. For definitions of MDL and litigation vocabulary used later in this guide, see the [mass tort glossary of MDL and litigation terms](/glossary).

## Best legal research platforms by need: quick picks

There is no single best legal research platform. The right pick depends on whether you need full-service research with a citator and AI (Westlaw or Lexis), news, research and practice tools in one subscription (Bloomberg Law), free or near-free access (your bar's Fastcase or Decisis plus free sources), or specialist layers for dockets, verdicts and contracts.

![Three-tier map of legal research platforms: enterprise platforms, low-cost and bar-benefit services, and free public sources](/images/blog-content/legal-research-tools/01-research-platform-tiers.jpg "The research market in three tiers. Most firms pair one paid or bar-benefit platform with free government and nonprofit sources.")

| Need | Pick | Why it fits | Pricing model (Oct 2026) |
|---|---|---|---|
| Full-service research with AI | Westlaw Edge with AI-Assisted Research or Advantage | KeyCite, Overruling Risk, AI Jurisdictional Surveys, Deep Research | Published: from $155.35/mo (new firms up to 10 attorneys, 3-year term) |
| Research plus practice guidance | Lexis+ with Protégé | Shepard's, Practical Guidance, Protégé AI | Custom quote; online store for 1 to 3 attorneys; 2-day trial |
| News, research and practice tools in one subscription | Bloomberg Law | News, primary and secondary sources and practice tools "for one price" (vendor wording) | Quote via demo |
| Free with bar dues (most states) | vLex Fastcase | Fastcase Library and Cert Citator through 80+ bars | Bar benefit |
| Free with bar dues (10 states) | Decisis | Case law, codes, regulations, dockets | Bar benefit or direct trial |
| Low-cost AI research | Descrybe | Primary-law search with the Cytator citator | $25 or $50 per month |
| Free case law | Google Scholar | State appellate since 1950, federal since 1923, SCOTUS since 1791 | Free |
| Free dockets and filings | CourtListener with RECAP | RECAP Archive, free opinions, alerts | Free (alert limits) |
| Official federal court records | PACER | Federal dockets and filings | $0.10/page, $3/document cap, waived at $30/quarter |
| Docket tracking at scale | CourtLink or Docket Alarm | Court dockets with alerts | Quote |
| Verdict and settlement research | Lexis Verdict & Settlement Analyzer | Verdict data on settlement patterns (vendor) | Quote |
| Contract and clause research | Law Insider (SimpleAI) | Clause and contract database | 7-day trial, then $1,790/user/year |

Prices verified October 22, 2026, on each vendor's site. Confirm current pricing with the vendor before you buy.

The sources behind each row, as of October 2026: Thomson Reuters lists [AI Jurisdictional Surveys on Edge with AI-Assisted Research and Deep Research on Advantage](https://legal.thomsonreuters.com/en/westlaw), says [KeyCite Overruling Risk](https://legal.thomsonreuters.com/en/products/westlaw/keycite) is available only on Westlaw, and publishes Advantage [from $256.75 a month for a single circuit](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-advantage/plans-pricing?segment=lawfirm). Those Westlaw prices apply only to new customers at firms with up to 10 attorneys. LexisNexis says Protégé is built into a platform that combines primary law, ["exclusive secondary sources, and Practical Guidance"](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page) with Shepard's. [Bloomberg Law](https://pro.bloomberglaw.com/) offers a demo rather than a price.

For the low-cost and free rows: [Fastcase](https://www.fastcase.com/) says it partners "with over 80 national, state, and county bar associations" as "a free member benefit." [Decisis](https://www.decisis.com/) positions itself for "bar associations or 1-2 attorney firms" and offers a trial. [Descrybe's pricing page](https://descrybe.com/pricing) lists its Platform at $50 a month and Legal Engine at $25. [Google Scholar's help page](https://scholar.google.com/intl/en/scholar/help.html) gives the coverage dates in the table, and [CourtListener's coverage page](https://www.courtlistener.com/help/coverage/) describes its archive and alerts. Docket Alarm is a [vLex product](https://vlex.com/) and CourtLink a LexisNexis one; both are quote-only.

### How these picks were chosen

The picks rest on documented features and published prices taken from each vendor's own pages on October 22, 2026. Ownership and product status come from press releases and help pages. Free sources are described from their own coverage pages. Where a fact comes from a secondary source, such as a law-library guide or a legal-tech news site, the sentence says so.

What the method does not include matters as much. There was no hands-on benchmark, no side-by-side test of search results and no input from vendors. No pick earns a commission, and no link is an affiliate link. The labels in the table are "best for a need," which is my judgment from the documented features, not a ranking of overall quality.

Picks also age. A vendor can reprice a plan, rename a product or drop a feature in a single quarter, and the market map below shows how often that has happened since 2023. That is why every price on this page carries a date and a reminder to confirm it with the vendor before buying.

### Picks for solo and small plaintiff firms

My starting point for a solo or small plaintiff firm is the free layer you already pay for through bar dues. Fastcase reaches members through more than 80 bars, and Duke Law's Goodson Law Library reports that every state bar gives members free access to vLex Fastcase or Decisis, with 10 states Decisis-only and Indiana and Wyoming offering both (table as of May 9, 2025). Add Google Scholar, CourtListener and PACER's free opinions, and many everyday questions are covered at no extra cost.

I would add a paid platform when citator depth, secondary sources or AI surveys start costing billable hours. Small firms can comparison-shop without a sales call: Westlaw publishes prices for new customers at firms with up to 10 attorneys, and Lexis lets 1 to 3 attorney firms view costs and buy through its online store.

A worked example. A two-lawyer car-accident practice mostly needs its home state's case law and statutes, one federal circuit for removal fights, and a citator it trusts. Westlaw's single-circuit tier is the published reference point here: Edge with AI-Assisted Research starts at $155.35 a month on a 3-year term, as of October 2026, for new customers at firms with up to 10 attorneys. Confirm with the vendor, and compare it against the bar benefit plus a Lexis store quote. The five-question test in the buying section helps decide whether the gap is worth paying for.

### Picks for mid-size and large litigation firms

Firms with more than 10 attorneys fall outside Westlaw's published pricing, so they negotiate, and the online prices serve only as a floor reference. The same is true at LexisNexis above the 1 to 3 attorney store tier and at every quote-only vendor.

The stack I would expect at this size has four pieces. One enterprise platform serves as the system of record, with its citator, secondary sources and AI layer. A second platform, or the bar benefit, gives a second citator opinion on the authorities that carry a motion. A docket tracker watches the cases the firm is in and the ones it is watching. Litigation analytics earn their cost when the firm appears before many judges in many venues.

Mass tort and multi-state dockets add two more layers: monitoring of the JPML and individual MDL dockets, and the regulatory and scientific databases behind product cases. The plaintiff-stack section covers both. For the organizational side of that growth, see [building research capacity for a multi-tort docket](/blogs/scaling-single-tort-to-multi-litigation).

## The 2026 market map: mergers, renames and retirements

Since 2021 the research market has consolidated. Fastcase absorbed Casemaker, merged with vLex and is now owned by Clio. Thomson Reuters bought Casetext and folded it into CoCounsel. LexisNexis renamed its AI platform. Westlaw moved to a three-version ladder, and FindLaw and the Caselaw Access Project changed hands.

| Date | Change | What it means for buyers |
|---|---|---|
| Jan 1, 2021 | Casemaker and Fastcase merge (per LawSites) | Old Casemaker logins move to Fastcase |
| Apr 4, 2023 | vLex and Fastcase merge (per LawSites) | One company behind both libraries |
| Aug 17, 2023 | Thomson Reuters closes Casetext purchase for $650M cash | Casetext product line ends |
| Mar 2024 | Caselaw Access Project restrictions expire | Search CAP cases in CourtListener |
| Dec 2, 2024 | Thomson Reuters sells FindLaw to Internet Brands | FindLaw no longer a TR site |
| Q1 2025 | Legacy Casetext users migrated to CoCounsel 2.0 | No standalone Casetext login |
| Jun 30, 2025 | Clio signs US$1B vLex deal | Practice software and research converge |
| Aug 5, 2025 | CoCounsel Legal and Westlaw Advantage launch | New top Westlaw tier |
| Nov 10, 2025 | Clio completes vLex acquisition | Fastcase, vLex, Docket Alarm under Clio |
| Feb 2026 | Lexis+ AI renamed Lexis+ with Protégé | Same product, new name |
| May 2026 | Law Insider legacy plans move to SimpleAI | Old Law Insider prices retired |

The dates come from the sources linked in the subsections below, plus [Law Insider's pricing page](https://www.lawinsider.com/pricing), which says "As of May 2026 — all legacy plans were upgraded into SimpleAI plans."

### Casetext became CoCounsel

Thomson Reuters [closed its acquisition of Casetext](https://www.thomsonreuters.com/en/press-releases/2023/august/thomson-reuters-completes-acquisition-of-casetext-inc.html) on August 17, 2023, "for a purchase price of $650 million in cash." The same release says Casetext was founded in 2013 and that its key product was CoCounsel, "an AI legal assistant powered by GPT-4." Today casetext.com redirects to [Thomson Reuters' CoCounsel page](https://www.thomsonreuters.com/en/cocounsel), and no standalone Casetext research product is offered.

According to Thomson Reuters' since-removed migration notice, legacy casetext.com CoCounsel 1.0 users were moved to CoCounsel 2.0 in the first quarter of 2025, in waves on February 1 and March 1, with both versions available until March 31, 2025. Any list that still quotes a Casetext monthly price is out of date. CoCounsel now sits inside Thomson Reuters' lineup alongside Westlaw, which the Westlaw section covers.

### Fastcase, vLex and Docket Alarm are now part of Clio

According to LawSites, [vLex and Fastcase announced a merger on April 4, 2023](https://www.lawnext.com/2023/04/in-major-legal-tech-deal-vlex-and-fastcase-merge-creating-a-global-legal-research-company-backed-by-oakley-capital-and-bain-capital.html), forming vLex Group. Clio, the practice management company, [signed a definitive agreement to acquire vLex](https://www.clio.com/about/press/clio-signs-definitive-agreement-to-acquire-vlex/) on June 30, 2025, in a deal "valued at US$1 billion," and completed it on November 10, 2025. The completion release names Clio Work, Clio Manage, Clio Grow and Clio Draft as the Clio products whose AI capabilities combine with vLex's Vincent AI.

The brands now say so themselves. Fastcase.com carries the banner "Fastcase is now part of Clio," and vLex's homepage is titled "vLex (Part of Clio)," with a product menu listing Vincent, vLex Library, Docket Alarm, vLex Labs and Fastcase Library. Two details matter for buyers. vLex says [Fastcase Library](https://vlex.com/fastcase-library) "is available exclusively through bar association partnerships," so most lawyers reach it through bar dues. The global [vLex Library](https://vlex.com/vlex-library), by vLex's account, gives "access to over one billion legal documents from 100+ countries."

### Lexis+ AI is now Lexis+ with Protégé

LexisNexis states on its product page that ["As of February 2026, Lexis+ AI® has been renamed Lexis+® with Protégé™."](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page) The company describes Protégé as "the AI assistant built into that platform," which combines primary law, "exclusive secondary sources, and Practical Guidance" with "Shepard's citation validation."

Readers who search for "Lexis+ AI" land on the renamed product, and the rename describes the same platform under a new name rather than a new purchase. In this guide, "Lexis+ AI" appears only as history, for example when Part 4 discusses a 2024 accuracy study that tested the product under its old name.

### Westlaw's version ladder ends at Advantage

According to LawSites, Thomson Reuters said [Westlaw Advantage "will now be the final version"](https://www.lawnext.com/2025/08/thomson-reuters-launches-cocounsel-legal-with-agentic-ai-and-deep-research-capabilities-along-with-a-new-and-final-version-of-westlaw.html) of Westlaw. LawSites traces the line from Westlaw to WestlawNext (2010), Westlaw Edge (2018), Westlaw Precision (2022) and Westlaw Advantage (2025). Thomson Reuters [launched CoCounsel Legal and Westlaw Advantage](https://www.thomsonreuters.com/en/press-releases/2025/august/thomson-reuters-launches-cocounsel-legal-transforming-legal-work-with-agentic-ai-and-deep-research) on August 5, 2025, and says "Deep Research on the new Westlaw Advantage is integrated into CoCounsel Legal."

The Westlaw page now offers three versions, each with a "View pricing" link: Westlaw Edge, Westlaw Edge with AI-Assisted Research and Westlaw Advantage. Precision and Classic are legacy names. They still appear in the page's own meta description, but the version cards list only the three current options, so those are the choices a buyer actually faces.

### FindLaw and the Caselaw Access Project changed hands

Thomson Reuters [closed its sale of FindLaw to Internet Brands](https://www.thomsonreuters.com/en/press-releases/2024/december/thomson-reuters-sells-findlaw-business-to-internet-brands.html) on December 2, 2024. Library guides and lists that still describe FindLaw as a Thomson Reuters site are stale on that point.

The Caselaw Access Project changed more quietly. Harvard's Library Innovation Lab [announced in March 2024](https://lil.law.harvard.edu/blog/2024/03/26/transitions-for-the-caselaw-access-project/) that the project's restrictions had expired, that "The Free Law Project already includes all CAP cases" in its CourtListener search engine, and that LIL keeps the bulk data at case.law while winding down its own search and API. The same post calls the project's original partner "Ravel Law, Inc. (now part of LexisNexis)." One last migration closes the loop: Fastcase says ["Casemaker users are being upgraded"](https://www.fastcase.com/casemaker/) to the Fastcase platform.

## How legal research works: a six-step workflow

Legal research follows the same sequence on every platform: frame the issue and jurisdiction, orient in secondary sources, find primary law, validate it with a citator, save it to the matter, and cite it with a record of what you checked.

1. Frame the issue, facts and jurisdiction.
2. Start in secondary sources.
3. Search primary law with the right syntax.
4. Validate every authority with a citator.
5. Save and organize research in the matter.
6. Cite, document and hand off.

![Six-step legal research workflow: frame the issue, secondary sources, primary law, citator check, organize, cite and record](/images/blog-content/legal-research-tools/02-six-step-research-workflow.jpg "A repeatable research workflow. The citator check is the step that free tools and AI answers most often skip.")

### 1. Frame the issue, facts and jurisdiction

Write the question as a one-sentence issue statement before you open any database. Then list the facts that could change the answer, the governing jurisdiction (state court, federal circuit or MDL transferee court) and the procedural posture. A motion to dismiss, a summary-judgment opposition and a jury-instruction fight each call for different authority.

A short template keeps the framing honest:

- Issue: the question in one sentence.
- Jurisdiction and court: state, federal district and circuit, or MDL transferee court.
- Key facts: the five or six facts that drive the answer.
- Posture and deadline: the motion or stage, and when the brief is due.
- Known authority: any case, statute or rule already in hand.

A PI example: a rear-end crash in a no-fault state, where the threshold question is whether the client's injury meets the state's serious-injury standard for suing in tort. Framing it that way sends you to the state's no-fault statute and the cases construing it, not to general negligence law.

### 2. Start in secondary sources

Secondary sources save time because they summarize the law and cite the primary authority behind each statement. Wex is careful to note that secondary authority is "never mandatory," so you read it to orient, then cite the cases and statutes it points to.

The main kinds are treatises on a subject, practice guides, law review articles, Restatements and pattern jury instructions, and CLE materials. Practical Law (Thomson Reuters) and Practical Guidance (LexisNexis) are vendor practice-guide products sold with or alongside the big platforms. Free options exist too, including law review articles indexed by Google Scholar and the annotations in Cornell's Wex.

One habit prevents most secondary-source errors: check the publication date and the latest supplement before relying on any statement. A treatise section written before a statute was amended, or before the state supreme court changed the rule, will point you confidently to law that no longer applies.

### 3. Search primary law with the right syntax

Every major platform offers two ways to search. Natural-language search reads a plain question and ranks results by relevance, which works well for orientation when you do not yet know the vocabulary. Terms and connectors (Boolean search) lets you specify exactly which words must appear and how close together. It is slower to write but precise and repeatable, which makes it the better choice for saved searches and alerts.

Thomson Reuters' [Westlaw Edge help page on terms and connectors](https://www.thomsonreuters.com/en-us/help/westlaw-edge/searching/search-with-terms-and-connectors/search-with-terms-and-connectors.html) defines the connectors below.

| Westlaw connector | What it does | Example |
|---|---|---|
| `!` | Root expander: words with multiple endings | `object!` finds object, objected, objection |
| `*` | Universal character inside a word | `withdr*w` finds withdraw and withdrew |
| `#` | Searches the term exactly as typed | `#damage` excludes damages |
| `/s` | Both terms in the same sentence | `design /s defect` |
| `+s` | First term precedes the second in the same sentence | `attorney +s fee` |
| `/p` | Both terms in the same paragraph | `hearsay /p utterance` |
| `+p` | First term precedes the second in the same paragraph | First term before second, same paragraph |
| `/n` | Terms within n words (1 to 255) | `personal /3 jurisdiction` |
| `%` | BUT NOT: excludes what follows | Drops results with the excluded term |
| `&` | AND: both terms appear | Use in Advanced Search |
| space (OR) | Either term appears | Use in Advanced Search |

Source: Thomson Reuters, Westlaw Edge help, "Search with Terms and Connectors," accessed October 24, 2026.

Thomson Reuters' page says to use Advanced Search for the AND (`&`) and OR (space) connectors. Lexis, Bloomberg Law and Fastcase use similar but not identical syntax, so check each platform's help page before you reuse a search string. Whatever the platform, apply filters before you read: narrow by jurisdiction, date range and court level first, so the results list matches the framing from step 1.

### 4. Validate every authority with a citator

A citator shows a source's subsequent history and the later documents that cite it, and flags negative treatment such as reversal or overruling. Thomson Reuters says KeyCite lets you ["verify whether a case, statute, regulation, or administrative decision is still good law ... and find citing references."](https://legal.thomsonreuters.com/en/products/westlaw/keycite) LexisNexis publishes a guide to the treatments its [Shepard's Signal indicators](https://www.lexisnexis.com/supp/lawschool/resources/be-confident-shepards.pdf) cover. vLex calls Fastcase Library's citator the Cert Citator, and Descrybe calls its own the Cytator. Part 2 explains the flags and signals. Free tools generally lack a full treatment citator, a gap the section on what free research costs covers.

Headnotes deserve a warning here. Wex says [headnotes](https://www.law.cornell.edu/wex/headnote) are "generally written by third party publishers" and "do not form a part of the judgment and are not binding precedents." They are useful for finding the right passage and the right point of law. Never quote a headnote as if it were the court's language.

The rule of thumb is simple: run the citator on every authority you cite, and run it again before filing. A case can be reversed between the day you draft and the day the brief goes out.

### 5. Save and organize research in the matter

Research that lives only in a browser history is lost the day the associate who did it leaves. Organize it by matter and by issue: one folder per matter, subfolders per issue, and a saved search with an alert for each issue that is still live. Every saved authority should carry a short note on why it matters (the holding, the pin cite, the fact pattern) so the next reader does not have to reread the case.

Export the final set into the matter file, whether that is a document management system or a case management platform. For how firms connect research folders to their matters, see [saving research to the matter in case management software](/blogs/legal-case-management-software). Firm-level brief banks and research memo libraries are covered in the operations section.

### 6. Cite, document and hand off

Cite to the official or preferred reporter for the court you are in, and confirm the pin cites against the opinion text rather than a headnote. Then keep a short research log, because "what did we check, and when?" is the first question a supervising lawyer, a client or a court may ask.

A research log template:

- Question: the issue statement from step 1.
- Sources and databases searched: platforms, free sites and secondary sources used.
- Search strings: the exact queries and filters.
- Key authorities and citator status: each authority with its flag or signal and the date checked.
- Open issues: what is unresolved and where to look next.

Hand the work off with a memo that separates holdings from commentary, so the reader can tell what the court decided from what a treatise or the researcher thinks it means. The workflow is the same on every platform, but platforms differ in content, citators, AI and price, starting with Westlaw.

## Westlaw: Edge, AI-Assisted Research and Advantage

Westlaw is Thomson Reuters' research platform, sold in [three versions](https://legal.thomsonreuters.com/en/westlaw): Westlaw Edge, Westlaw Edge with AI-Assisted Research, and Westlaw Advantage, which adds agentic Deep Research. Its citator is KeyCite. Published online prices start at $155.35 a month, but only for new customers at firms with up to 10 attorneys.

### The three Westlaw versions compared

Each step up the Westlaw ladder adds AI capability on top of the one below it. Thomson Reuters says Westlaw Advantage "includes all the features and functionality of Westlaw Edge with AI-Assisted Research," so the versions stack rather than branch. The question for a buyer is how far up the stack the firm's work justifies going.

| Version | What Thomson Reuters lists | Published starting price (Oct 2026) | Term |
|---|---|---|---|
| Westlaw Edge | Missed-authority report on uploaded documents; analytics on judges, courts, attorneys, firms, damages and case types; KeyCite Overruling Risk; statute version compare | Not published in page text; use the plan builder | Multi-year discounts |
| Westlaw Edge with AI-Assisted Research | AI-Assisted Research, AI Jurisdictional Surveys, AI-enhanced Quick Check, WestSearch Plus | $155.35/mo single circuit; $266.50/mo all states and federal | 3-year term |
| Westlaw Advantage | Everything in Edge with AI-Assisted Research, plus Deep Research, Claims Explorer, Litigation Document Analyzer | $256.75/mo single circuit; $399.75/mo all states and federal | Not stated; multi-year discounts |

Prices verified October 22, 2026, on Thomson Reuters' plan pages. Online prices apply to new customers at firms with up to 10 attorneys. Confirm with the vendor.

The Edge feature list is the base every version shares. Thomson Reuters describes a report that surfaces authority that "may have been missed or is contrary to your opponent's position" after you upload a document, "data-driven insights on judges, courts, attorneys, law firms, damages, and case types," and a tool to "instantly compare any two versions of a statute or regulation." The AI-Assisted Research tier adds natural-language research, AI Jurisdictional Surveys, an "AI-enhanced Quick Check" and WestSearch Plus. Advantage adds the "agentic AI of Deep Research," Claims Explorer, which Thomson Reuters says helps you "find the strongest claims for your legal matter," and the Litigation Document Analyzer.

If you have used Westlaw before, the names may not match your memory. Westlaw Precision and Westlaw Classic still appear in the meta description of Thomson Reuters' Westlaw page, but the page itself offers only the three versions above, so treat Precision and Classic as legacy names. Thomson Reuters also told LawSites that Advantage "will now be the [final version](https://www.lawnext.com/2025/08/thomson-reuters-launches-cocounsel-legal-with-agentic-ai-and-deep-research-capabilities-along-with-a-new-and-final-version-of-westlaw.html)" of Westlaw, with continuous updates after it rather than another renamed release.

### KeyCite flags and Overruling Risk

[KeyCite](https://legal.thomsonreuters.com/en/products/westlaw/keycite) is Westlaw's citator. Thomson Reuters says it verifies "whether a case, statute, regulation, or administrative decision is still good law," checks whether a patent or trademark is still valid, and finds citing references. Its main features are Citing References (sorted with a "depth of the discussion" indicator), History, and KeyCite Alerts, which notify you when the status of a saved authority changes.

![Explainer of citator status flags: red for no longer good law, yellow for negative history, blue stripes for appealed](/images/blog-content/legal-research-tools/03-citator-flags-explainer.jpg "How KeyCite status flags read, per Thomson Reuters' help page. Shepard's and other citators use their own signals and wording.")

Thomson Reuters' [KeyCite status flags help page](https://www.thomsonreuters.com/en-us/help/drafting-assistant/westcheck/keycite-status-flags) defines the case flags this way:

- Red flag: the case is no longer good law for at least one point of law.
- Yellow flag: the case has some negative history but has not been reversed or overruled.
- Blue-striped flag: the case has been appealed to the U.S. Courts of Appeals or the U.S. Supreme Court (appeals to agencies are excluded).
- Overruling Risk icon: the case relies on a case or proposition that was explicitly overruled.
- Blue H: the case has some history.
- Green C: the case has citing references but no direct or negative history.

Statutes get flags too. A red flag on a statute means it was amended by a recent session law or rule, repealed, superseded, held unconstitutional or preempted in whole or in part; a yellow flag covers cautions such as renumbering or transfer by a recent session law.

Overruling Risk is the feature Thomson Reuters describes as "available exclusively on Westlaw." It "warns you when a point of law has been implicitly undermined based on its reliance on an overruled or otherwise invalid prior decision," and Thomson Reuters says AI "identifies bad law that has no direct citations pointing to its invalidity." In practice, picture an appellate decision from 2012 that rested its holding on a 1998 precedent. The 1998 case is overruled in 2023. No court has yet cited the 2012 decision negatively, so a traditional citator shows nothing alarming. Overruling Risk is meant to catch that gap.

Treat every flag as a starting point. A red flag can concern a point of law your brief never touches, and a clean green C does not prove the reasoning still holds. Open the negative citing reference, read the relevant passage, and decide for yourself.

### CoCounsel Legal, Deep Research and Practical Law

Casetext's CoCounsel assistant now lives inside Thomson Reuters' product line; casetext.com redirects to [Thomson Reuters' CoCounsel page](https://www.thomsonreuters.com/en/cocounsel). Thomson Reuters [launched CoCounsel Legal on August 5, 2025](https://www.thomsonreuters.com/en/press-releases/2025/august/thomson-reuters-launches-cocounsel-legal-transforming-legal-work-with-agentic-ai-and-deep-research), and its release says "Deep Research on the new Westlaw Advantage is integrated into CoCounsel Legal."

Thomson Reuters describes Deep Research as "grounded in Westlaw and Practical Law content," and says it "explains its process, sources its answers" and works "with human oversight." Those are vendor descriptions. They tell you what the tool is designed to do, not how often it gets the answer right. The evidence on AI research accuracy, including the published studies and the court sanctions cases, is covered in the AI section later in this guide.

The Westlaw page lists CoCounsel Legal, CoCounsel Essentials and Practical Law as related offerings. None of them had a published price on the pages read for this guide, so treat them as quote-based.

Practical Law deserves a separate note because it solves a different problem. It is a secondary-source and know-how library: Thomson Reuters says it is backed by "650+ attorney editors" and "thousands of expertly maintained how-to guides, templates, checklists." For a litigator stepping into an unfamiliar procedural area, a practice note or checklist often gets you to the right statutes and leading cases faster than a cold keyword search. Whether that is worth a second subscription depends on how often your firm works outside its core practice.

### Westlaw pricing, terms and who qualifies

Thomson Reuters publishes online prices for two of its three versions. On the [Edge with AI-Assisted Research plans page](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge-aar/plans-pricing?segment=lawfirm), the plan "starts at $155.35 per month for single-circuit coverage with a 3-year term," and all states and federal "starts at $266.50 per month with a 3-year term." Thomson Reuters adds that "all plans include AI-Assisted Research and AI Jurisdictional Surveys." The [Westlaw Advantage plans page](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-advantage/plans-pricing?segment=lawfirm) says Advantage "starts at $256.75 per month for single-circuit coverage and $399.75 per month for all states and federal coverage," and that firms "may also qualify for lower pricing with multi-year subscription discounts." It does not name the term behind those figures.

Coverage drives the price. Thomson Reuters says "a single-circuit plan includes state and federal primary law for one circuit," while the all states and federal plan adds coverage across all states, federal courts and "additional legislative history resources." A single-state PI practice that rarely leaves its circuit may be well served by the narrower plan. A firm filing in several circuits, or doing multistate mass tort work, will feel the limit quickly.

Term discounts are spelled out. For Edge with AI-Assisted Research, "2-year plans include a 12% discount, and 3-year plans include an 18% discount off the base monthly price." Advantage uses the same structure: save 12% with a 2-year subscription and 18% with a 3-year one. Because Thomson Reuters ties the $155.35 figure to a 3-year term, that published starting price already reflects the 18% discount.

Worked arithmetic, as of October 2026 and before taxes or any out-of-plan charges: 36 months at $155.35 is a $5,592.60 commitment, and 36 months at $266.50 is $9,594.00. No equivalent math is given for Advantage here, because its page does not state the term.

Eligibility is narrow. Thomson Reuters says "online pricing is available for new customers at law firms with up to 10 attorneys" on its [Westlaw Edge plans page](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge/plans-pricing?segment=lawfirm); larger firms go through sales, and existing customers are not eligible. Most new customers get access "within 1 business day." The same page lists "Free trials & demos" without stating a trial length, and it shows no price for base Edge in its page text, so this guide quotes none. Before signing, confirm the numbers with Thomson Reuters and read the order form for auto-renewal language and out-of-plan charges; the contract checklist later in this guide lists what to ask.

### Who Westlaw fits best

My view, based on the documented features rather than hands-on testing: Westlaw fits firms that want KeyCite and litigation analytics in the same subscription. The Edge analytics on judges, courts, opposing counsel and damages are useful when you are valuing a case or deciding where to file. If you do that work, pair damages analytics with [state-specific case value math](/tools/mva-case-value-estimator) rather than relying on either alone.

It also fits litigators who regularly need AI Jurisdictional Surveys across states, and firms of up to 10 attorneys that want to see a published price before talking to a sales rep. Westlaw is a weaker fit when budget is the binding constraint and your state bar's free research benefit already covers your home jurisdiction well. In that case the paid subscription mostly buys analytics and AI, and you should price those specifically.

## Lexis+ with Protégé and the LexisNexis research stack

LexisNexis's flagship research platform is Lexis+, and Lexis+ with Protégé adds generative AI on top of it. LexisNexis says that "as of February 2026, [Lexis+ AI® has been renamed Lexis+® with Protégé™](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page)." The citator is Shepard's. Pricing is customized, firms of 1 to 3 attorneys can buy online, and LexisNexis advertises a 2-day free trial.

Prior Lexis+ AI users have access to the renamed product, so if a colleague mentions "Lexis+ AI," they mean the same line. The sections below cover the two plans, the citator, the AI configuration, and how buying works for small firms.

### Lexis+ versus Lexis+ with Protégé

The clearest distinction comes from the [LexisNexis online store FAQ](https://store.lexisnexis.com/lawfirms): "Lexis+ is a premium plan that integrates Practical Guidance, analytical tools, and Document Analysis… Lexis+ with Protégé includes all of Lexis+ and integrates generative AI." In other words, the AI plan is a superset, not a separate product.

LexisNexis describes Lexis+ with Protégé as "a legal AI solution built for drafting, research, and analysis," and Protégé as "the AI assistant built into that platform." The platform combines primary law, "exclusive secondary sources, and Practical Guidance" with "Shepard's citation validation," in LexisNexis's words.

Practical Guidance is LexisNexis's know-how library, the counterpart to Thomson Reuters' Practical Law. LexisNexis says it offers "expert-authored practice notes, annotated forms, and drafting guidance across 25 practice areas."

Which plan to trial first depends on one question: does the firm want AI drafting and research inside the research platform itself? If yes, trial Lexis+ with Protégé, since it includes everything in Lexis+. If the firm already uses a separate AI tool, or wants to keep AI out of research until it has a verification policy, Lexis+ is the cleaner comparison against a non-AI Westlaw Edge plan.

### Shepard's signals and what they tell you

Shepard's is LexisNexis's citator. A [LexisNexis law-school guide](https://www.lexisnexis.com/supp/lawschool/resources/be-confident-shepards.pdf) (© 2023) lists Shepard's Signal indicators "for seven treatments":

1. "Strong negative analysis impacting your decision"
2. "Strong negative analysis impacting your statute"
3. "Analysis questions precedential value of your decision"
4. "Possible negative analysis impacting your case"
5. "Case has been followed by other cases and courts"
6. "Analysis is available that is not positive nor negative"
7. "Citing references are available without analysis"

The guide is the source for the wording only. Check the current Shepard's help page for how each signal appears on screen.

Here is how a litigator can read them. A "strong negative" signal means stop and read the citing decision before relying on the authority at all. "Questions precedential value" means a later court has cast doubt without overruling, so find out why and whether that court binds yours. "Possible negative" is a prompt to look, not a verdict. A "followed" signal is reassurance that courts keep relying on the case; it is not proof the case supports your specific proposition. The neutral and no-analysis signals simply tell you citing references exist.

KeyCite expresses similar judgments through flags and an Overruling Risk warning, while Shepard's uses treatment-based signals. Neither approach replaces reading the cases.

### Protégé Vaults, model choices and uploads

Protégé lets you upload documents and work with them, and the limits shape real workflow. LexisNexis says you can "create up to 50 Vaults with 1-500 documents per Vault," that "uploading 11 or more documents automatically prompts you to create a Vault," and that uploads of up to 10 documents not saved to a Vault "are automatically purged at the end of each session."

That last rule matters in practice. Say a paralegal uploads eight deposition excerpts to ask Protégé for inconsistencies, closes the browser at 6 p.m., and comes back the next morning. The excerpts are gone. For any document set you expect to revisit, such as a medical chronology, the defendant's discovery responses, or a set of expert reports, create a Vault on day one and name it by matter.

LexisNexis offers two model configurations. "Protégé Legal AI" is a multi-model setup that "grounds all responses in LexisNexis sources and includes Shepard's citations." "Protégé General AI includes access to large language models from OpenAI, Google, and Anthropic." A sensible working rule: use the grounded Legal AI option when you need authority you might cite, and treat General AI output as drafting help that still needs every proposition checked against primary law.

Uploading client documents raises confidentiality questions under the ethics rules. Those are covered with the ABA guidance in the AI section later in this guide.

### Lexis pricing, the online store and the trial

LexisNexis does not publish a list price. Its product page says pricing "varies based on factors such as the size of your organization, specific capabilities required, and the scope of content access. LexisNexis offers customized pricing." So the three levers are firm size, features and content scope, which mirrors Westlaw's coverage-based approach without the published numbers.

Small firms have a self-serve route. LexisNexis's [small-law-firm page](https://www.lexisnexis.com/en-us/small-law-firms.page) says "1-3 attorney law firms can view costs and buy products directly from the LexisNexis Online Store," and order processing "can take up to 72 hours." Store prices load by script and were not captured for this guide, so it quotes none. Check the store yourself, as of the day you buy, and treat any third-party "starting at" figure for Lexis with suspicion unless it links to LexisNexis.

Both the product page and the small-firm page advertise a 2-day free trial. Two days is short. Plan the trial before you start the clock, using the side-by-side method below, so you test real questions rather than clicking around.

### Who Lexis fits best

My view, again from documented features: Lexis fits firms that want Shepard's and Practical Guidance in one place, especially those doing drafting-heavy work where annotated forms save time. It suits small firms of 1 to 3 attorneys that would rather buy online than sit through a sales call.

It also suits litigation teams that will use the rest of the [LexisNexis research stack](https://www.lexisnexis.com/en-us/solutions/legal-research.page): CourtLink for dockets, Lexis Verdict & Settlement Analyzer, Lexis Expert Research On-Demand, Context and State Net, which are covered in the specialist-tools section of this guide. LexisNexis's analytics product [Lex Machina](https://lexmachina.com/) is now, in its own words, "empowered by Protégé," which matters if you want analytics and AI from one vendor.

## Westlaw vs LexisNexis: which fits your firm

Neither platform is better for every firm. Both offer full primary law, a mature citator, editorial secondary sources and an AI layer; the practical differences are the citator's signals, the AI products, the analytics and practice-guidance add-ons, and how each vendor sells to small firms.

### Where the two platforms differ for litigators

| Factor | Westlaw | Lexis+ with Protégé |
|---|---|---|
| Citator | KeyCite flags plus Overruling Risk | Shepard's seven treatment signals |
| AI research layer | AI-Assisted Research, AI Jurisdictional Surveys; Deep Research on Advantage | Protégé Legal AI grounded in Lexis sources; Protégé General AI option |
| Practice guidance | Practical Law, a related offering | Practical Guidance, 25 practice areas |
| Analytics | Judge, court, attorney, firm and damages insights in Edge | Lex Machina and Context, sold separately |
| Document upload | Missed-authority report; Litigation Document Analyzer on Advantage | Vaults: up to 50, each with 1 to 500 documents |
| Published small-firm price | From $155.35/mo, new firms up to 10 attorneys | Online store for 1 to 3 attorneys; prices not captured |
| Trial | "Free trials & demos", length not stated | 2-day free trial |

Features and prices as listed by each vendor, verified October 22, 2026.

A few rows deserve comment. On analytics, Westlaw bundles judge and damages insights into Edge, while LexisNexis sells its heavier analytics as separate products; compare total cost, not the base plan. On documents, the two vendors solve different problems: Westlaw's tools check your brief or the other side's filing for missing or contrary authority, while Protégé Vaults are a workspace for asking questions across a document set. On pricing transparency, Westlaw publishes numbers for firms of up to 10 attorneys, and Lexis lets firms of up to 3 attorneys buy online.

### How to run a fair side-by-side trial

1. Pick five real research questions from closed matters in your jurisdiction, ones where you already know which authorities mattered.
2. Run each question on both platforms with the same time budget, for example 20 minutes per question per platform.
3. Record which authorities each platform found and which of them you would actually cite.
4. Run both citators on the same five key cases and compare the signals each one returns, then read the negative citing references behind any warning.
5. Test the AI layer on one question you already know the answer to, and check every citation it returns; the AI section later in this guide gives a full verification protocol.
6. Before the trial ends, ask both vendors for the order form, renewal terms and any out-of-plan charges, in writing.

Because the Lexis trial lasts only 2 days, schedule it last. Run your Westlaw trial or demo first, finalize the five questions and your scoring sheet, and then start the Lexis clock with everything ready.

## Bloomberg Law, vLex Fastcase and other paid platforms

Beyond Westlaw and Lexis, the paid options that matter are Bloomberg Law (research plus news and practice tools in one subscription), vLex with Vincent AI and Fastcase (all part of Clio since November 2025), Decisis, Descrybe for low-cost AI research, and Law Insider for contract and clause work.

These platforms differ more from one another than Westlaw and Lexis differ from each other. Some are full research services, some are bar-benefit databases, and one is a clause library. The sections below say what each vendor documents and where you will have to ask.

### Bloomberg Law

[Bloomberg Law](https://pro.bloomberglaw.com/) positions itself as "trusted news, primary and secondary sources, practice tools, and expert legal analysis" in one place, "for one price." It points to "40+ news channels and the Bloomberg Terminal" as part of what sets it apart, which makes it a natural fit for practices where regulatory and business news drive the legal work. Bloomberg's page also says Bloomberg Industry Group "has acquired Regology," which it calls "an AI-powered regulatory intelligence platform"; the page gives no date for the deal.

There is no published price. The buying path is a "Request Demo" form, with firm-size options running from Solo to 50+ practitioners.

The pages available for this guide did not document Bloomberg Law's dockets, litigation analytics or citator. Ask about all three in the demo, and ask for the citator's signal definitions in writing so you can compare them with KeyCite and Shepard's. Ask, too, what the "one price" includes and what costs extra.

### vLex, Vincent AI and Fastcase under Clio

The lineage is short. LawSites reported that vLex and Fastcase announced their [merger on April 4, 2023](https://www.lawnext.com/2023/04/in-major-legal-tech-deal-vlex-and-fastcase-merge-creating-a-global-legal-research-company-backed-by-oakley-capital-and-bain-capital.html), and [Clio completed its US$1 billion vLex acquisition on November 10, 2025](https://www.clio.com/about/press/clio-completes-landmark-1b-vlex-acquisition-series-g-5b-valuation/). [Fastcase's own site](https://www.fastcase.com/) now says "Fastcase is now part of Clio."

The [vLex site](https://vlex.com/) lists five products: Vincent, vLex Library, Docket Alarm, vLex Labs and Fastcase Library. Clio's release places them alongside its own products, Clio Work, Clio Manage, Clio Grow and Clio Draft, and calls the combined offering the "Intelligent Legal Work Platform." For how the practice management side fits with vLex, see [how Clio's practice management suite fits with vLex](/blogs/law-practice-management-software).

[vLex Library](https://vlex.com/vlex-library) includes Precedent Map, which uses "color-coded visual indicators" to show which precedents have been "positively cited, criticized, or overruled," and Key Passages, which "surfaces the most influential sections of legal decisions" using "citation frequency analysis." [Fastcase Library](https://vlex.com/fastcase-library) has its own citator, Cert Citator, which vLex says "instantly reveals when cases have been overturned or criticized," plus AI Case Analysis and an Interactive Timeline.

Access is where buyers get confused. vLex says Fastcase Library "is available exclusively through bar association partnerships," serving "over 1 million lawyers nationwide through 80+ state, county, and specialty bar associations." You cannot subscribe to it directly; most US lawyers meet it as a bar benefit, and the free-sources section of this guide has the state table. Fastcase's site similarly says it partners "with over 80 national, state, and county bar associations" as "a free member benefit." Lawyers who want to buy look at vLex and Vincent instead, which offer a "Start a Free Trial" option and quote-based plans.

On scale, vLex says it holds "over one billion legal documents from more than 100 countries." Clio's release describes Vincent as drawing "from vLex's library of more than one billion editorially enriched documents across 110 jurisdictions." The two figures use different units, so quote whichever one you need with its source. vLex also lists Docket Alarm with "850+ million court records," U.S. content from "250+ publishers" with "daily updates across 400+ courts," and "SOC2 and ISO 27001 certified" security. These are vendor claims.

For litigators, two Vincent features stand out. vLex says Vincent can "generate 50-state surveys instantly," a direct competitor to Westlaw's AI Jurisdictional Surveys. And on January 12, 2026, Clio announced Vincent Studio, a "no-code workflow builder" that is "available for Enterprise customers."

### Decisis

[Decisis](https://www.decisis.com/) describes itself as "a legal research database at a cost-effective price point" and calls itself "the clear choice for bar associations or 1-2 attorney firms." That is vendor positioning. The content it lists is "federal and state case law, codes, regulations and select agency materials," plus "forms, dockets, and news."

Most lawyers who use Decisis get it through their bar. According to a [Duke Law library guide](https://law.duke.edu/lib/statebarassociations), Decisis is the bar benefit in 10 states and one of two options in Indiana and Wyoming; the full state table is in the free-sources section of this guide.

For direct buyers, the site invites you to "TRY BEFORE YOU BUY" and "TAKE A TRIAL," but shows no price. Ask for the price, the term and whether dockets and forms are included in the base plan. If your bar already provides Decisis, run the trial against your bar access first to see what, if anything, a paid plan adds.

### Descrybe

Older articles describe Descrybe as a free AI legal search tool. That is out of date. The original site says Descrybe ["has moved to Descrybe.com"](https://descrybe.ai/) and "has since expanded into a paid legal research platform."

[Descrybe's pricing page](https://descrybe.com/pricing) lists two plans as of October 2026. The Descrybe Platform costs $50 a month and includes the Legal Engine with "150 full research runs per month." The Descrybe Legal Engine alone costs $25 a month, which Descrybe describes as a "flat monthly price with no per-search charges." Both renew monthly, and Descrybe says "payments are non-refundable." A launch code, LAUNCH50, takes 50% off the first month. The Legal Engine works inside ChatGPT, Claude, Microsoft 365 Copilot and Perplexity, and through an "Open Connector." Academic pricing runs through NELLCO.

On its [homepage](https://descrybe.com/), Descrybe says its tools "research primary law, resolve citations, verify quotations, check subsequent treatment, and retrieve source text." Its citator is called the "Cytator."

My view: Descrybe suits solos and small firms that want AI-assisted primary-law research at a published monthly price, with a citator check built in, and that do not need the editorial secondary sources of the large platforms. Because renewals are monthly and non-refundable, test it for a month on live matters before relying on it, and check its citations as you would any AI output.

### Law Insider for contract and clause research

Clause research means finding how other drafters have written a specific provision, then adapting the language. Litigators need it more often than they expect: settlement agreements, releases, lien resolution language, confidentiality provisions, co-counsel and referral agreements, and vendor contracts all benefit from seeing a range of real examples before you draft.

[Law Insider](https://www.lawinsider.com/pricing) says that "as of May 2026, all legacy plans were upgraded into SimpleAI plans," and the product is now "powered by SimpleDocs." The pricing page shows a free 7-day trial, then $1,790 per user per year on the annual view (which advertises savings of up to 16%). An Enterprise tier is custom-priced, and limited free access to the database remains.

Law Insider says SimpleAI includes "access to 20M+ contract & 200M+ clauses" and a Word integration, and that it is "used by more than 10,000 organizations." These are vendor claims. Older monthly and day-pass prices quoted elsewhere no longer appear on the pricing page, so ignore them. For a PI firm, the question is volume: if you draft releases and lien language a few times a year, the free access may be enough; if you draft them weekly, the trial will show whether the clause search saves real time.

### Paid platforms compared side by side

| Platform | Owner | Citator | AI layer | Dockets | Price model (Oct 2026) | Best for (my view) |
|---|---|---|---|---|---|---|
| Westlaw | Thomson Reuters | KeyCite | AI-Assisted Research; Deep Research on Advantage | Not covered on pages read | From $155.35/mo for new firms up to 10 attorneys; quote above that | Full-service litigation research |
| Lexis+ with Protégé | LexisNexis | Shepard's | Protégé Legal AI and General AI | CourtLink, sold separately | Custom; online store for 1 to 3 attorneys | Research plus practice guidance |
| Bloomberg Law | Bloomberg Industry Group | Ask in demo | Ask in demo | Ask in demo | Quote via demo | Research with news |
| vLex and Vincent | Clio | Precedent Map | Vincent AI | Docket Alarm | Free trial, then quote | Multi-state and global research |
| Fastcase Library | Clio | Cert Citator | AI Case Analysis | Via Docket Alarm (separate) | Bar benefit only | Bar members on a budget |
| Decisis | Decisis | Not listed on homepage | Not listed | "Dockets" listed | Bar benefit or trial | Bar members in Decisis states |
| Descrybe | Descrybe | Cytator | Built-in AI search | No | $25 or $50/mo | Low-cost AI research |
| Law Insider | SimpleDocs | No | SimpleAI | No | $1,790/user/yr after 7-day trial | Contract and clause research |

Compiled from vendor pages on October 22 and 24, 2026. "Ask in demo" means the vendor page did not document it. Confirm every item with the vendor.

Read the table by column, not by row. If a citator is non-negotiable, three platforms leave you asking or have none listed. If price transparency matters, only Westlaw, Descrybe and Law Insider publish numbers. And if your bar already gives you Fastcase or Decisis, the real question is what a paid platform adds on top. Paid platforms are only half the market, because a great deal of reliable law is free.

## Free legal research sites worth using

The free sources worth using are Google Scholar and CourtListener for case law, Cornell LII for statutes, regulations and definitions, govinfo for authenticated federal text, the eCFR and Federal Register for regulations, and PACER's free opinions plus the RECAP Archive for federal court records. None of them includes a full treatment citator, so verify every authority before you cite it.

| Source | Run by | What it covers | Citator or alerts | Official status | Best use |
|---|---|---|---|---|---|
| Google Scholar | Google | US state appellate cases since 1950, federal since 1923, Supreme Court since 1791 | "Cited by" links only | Unofficial, no warranty | Quick case-law checks |
| CourtListener | Free Law Project | Opinions back to the 1650s, RECAP filings, oral arguments | Free alerts with limits | Unofficial | Dockets and free opinions |
| case.law | Harvard Library Innovation Lab | CAP bulk data and readable cases | None | Unofficial | Historical bulk research |
| Cornell LII | Cornell Law School | U.S. Code, CFR, Wex | None | Unofficial | Statutes and definitions |
| Justia | Justia | Free case law and codes | None | Unofficial | Quick lookups |
| FindLaw | Internet Brands since Dec 2024 | General legal information | None | Unofficial | Orientation |
| govinfo | GPO | Federal Register, CFR and other collections | None | Authenticated signed PDFs | Official federal text |
| eCFR | Office of the Federal Register and GPO | CFR updated daily | None | Editorial compilation, not yet official | Current regulations |
| PACER | Federal judiciary | Federal dockets and filings | None | Official court records | Federal dockets |

Coverage as described on each source's own pages, October 22, 2026.

### Google Scholar case law

Google Scholar is the fastest free way to pull a known case or skim how courts have read a doctrine. On the search page, choose "Case law," then "Select courts" to limit results to a state or federal circuit. According to [Google Scholar's help page](https://scholar.google.com/intl/en/scholar/help.html), its case-law coverage is "published opinions of US state appellate and supreme court cases since 1950, US federal district, appellate, tax and bankruptcy courts since 1923 and US Supreme Court cases since 1791."

Those dates leave real holes. State appellate decisions before 1950 are outside the stated scope, and state trial-court orders, which matter a great deal in personal injury practice, are largely missing. The "Cited by" link lists later cases that cite an opinion, but it shows no treatment signal, so you have to open and read each citing case to learn whether it followed, distinguished or rejected the point.

Google is candid about the limit. The same help page says legal opinions "are provided for informational purposes only" and that "Google does not warrant that the information is complete or accurate." Treat Scholar as a first look and a cross-check, never as the final validation step before filing.

### CourtListener and the RECAP Archive

CourtListener is the most useful free source for federal court records because it pairs an opinion search engine with millions of filings that users have bought from PACER. It is run by the [Free Law Project](https://free.law/about/), which describes itself as a "501(c)(3) nonprofit" "Started in 2010" and also runs the RECAP Suite and Bots.law.

[CourtListener's coverage page](https://www.courtlistener.com/help/coverage/) says "Our archives go back to the 1650's" and lists "Tens of millions of federal filings in the RECAP Archive," oral argument recordings, judges' financial disclosure records, and state and Supreme Court filings.

RECAP is the clever part. According to [the RECAP project page](https://free.law/recap), you install the browser extension (Chrome, Edge, Safari or Firefox), and every document you buy in PACER is added to the public archive. In return, "Anything somebody else has added to the archive will be available to you for free — right in PACER itself." The archive includes "every free opinion in PACER" and is uploaded to the Internet Archive.

Free Law Project also offers docket and search alerts, free up to a limit, with higher limits for extension users and paying members. Attorneys can route their ECF notices to an "@recap.email" address so the free copy that comes with each notice also lands in the archive.

For a plaintiff firm, the payoff is following a large federal proceeding without paying for every page. See [how one MDL's public filings are followed](/blogs/social-media-addiction-lawsuit) for an example of the kind of docket that free archive supports.

### The Caselaw Access Project after 2024

The Caselaw Access Project (CAP) is no longer a search engine. It is now a bulk-data and readable-case site, and its cases are searchable in CourtListener instead.

CAP was Harvard's effort to digitize American case law. On its [project page](https://lil.law.harvard.edu/our-work/caselaw-access-project/), the Harvard Library Innovation Lab (LIL) says that from 2013 to 2018 it "digitized over 40 million pages" with Ravel Law into "a dataset of over 6.7 million cases" covering "360 years" of US legal history. The CAP API and bulk data launched on October 31, 2018. LIL's own pages round the count differently, so quote one figure with its source rather than a number from a list article.

In March 2024 the project changed shape. LIL's [transition announcement of March 26, 2024](https://lil.law.harvard.edu/blog/2024/03/26/transitions-for-the-caselaw-access-project/) said CAP's access limitations "expired this month," that "The Free Law Project already includes all CAP cases ... in its CourtListener search engine," and that LIL was "winding down services ... such as the search function and API." The old site was to stay up until September 2024.

What remains at case.law is bulk data and readable case pages. That is useful for researchers who want whole jurisdictions as data, or a historical opinion in clean text. For searching, go to CourtListener. Lists that still describe CAP as a live search tool, or give it a precise case count without a source, are out of date.

### Cornell LII and Wex

Cornell's Legal Information Institute (LII) is the best free place to read a federal statute or regulation in clean, linked text and to look up a legal term. Its tagline is "Free Law For All," and it hosts the U.S. Code, the Code of Federal Regulations and Wex.

LII's [U.S. Code index](https://www.law.cornell.edu/uscode/text) runs from Title 1 to Title 54. Title 53 is "[RESERVED]" and Title 54 is "National Park Service and Related Programs," a detail that confuses people who expect a continuous run. Each section page links its definitions and cross-references, which makes LII faster to read than a PDF.

[Wex](https://www.law.cornell.edu/wex) is "a free legal dictionary and encyclopedia sponsored and hosted by the Legal Information Institute [LII] at Cornell Law School." According to [LII's about page](https://www.law.cornell.edu/lii/about/about_lii), "Cornell Law students and recent graduates research and draft the content" under LII's supervision. Wex says plainly that "Nothing in Wex should be construed as legal advice."

Best use: reading a statute's current text, checking a definition, and orienting yourself in an unfamiliar federal area. LII copies are unofficial. When you need text a court will treat as authoritative, pull it from govinfo, covered two sections below.

### Justia, FindLaw and other free aggregators

Free aggregators are convenient front doors, not sources of record. Justia publishes free case law, state and federal codes, and an attorney directory. FindLaw offers plain-English legal information for consumers along with a lawyer directory. Both are quick ways to orient yourself or find a citation you can then pull from an official source.

Ownership matters for understanding what you are reading. Thomson Reuters [closed the sale of FindLaw to Internet Brands on December 2, 2024](https://www.thomsonreuters.com/en/press-releases/2024/december/thomson-reuters-sells-findlaw-business-to-internet-brands.html). Some research guides have not caught up; for example, [Temple University's library guide](https://guides.temple.edu/c.php?g=1046977&p=7597201), last updated July 25, 2025, still describes FindLaw as hosted by Thomson Reuters. That is a small error, but it shows why any list of research sources needs a date on it.

Two government and nonprofit sites belong in the same toolkit. Congress.gov publishes federal bill text, status and legislative history materials, which helps when you need to know what Congress meant or when a provision took effect. Oyez offers audio and summaries of Supreme Court oral arguments.

The rule for every aggregator is the same. Use it for speed, then check that the text is current and pull the official version before you cite it. Aggregators update on their own schedules, and none of them flags a case that has been overruled.

### Govinfo, the eCFR and the Federal Register

For federal text, three levels of reliability matter: official, authenticated and convenience copies. Govinfo gives you authenticated official text; the eCFR gives you a current but unofficial compilation; LII and the aggregators give you convenient reading copies.

The Government Publishing Office says on its [govinfo authentication page](https://www.govinfo.gov/about/authentication) that "GPO uses digital signature technology to provide evidence of authenticity and integrity." Signed PDFs display an eagle logo reading "Authenticated U.S. Government Information," and the Federal Register "and other collections" are signed. If a judge or opponent questions your text, a signed govinfo PDF settles it.

The Federal Register itself carries legal weight. Under [44 U.S.C. § 1507](https://www.law.cornell.edu/uscode/text/44/1507), "The contents of the Federal Register shall be judicially noticed," and the Register may be cited by volume and page. That is why proposed and final rules, preambles and agency notices should be cited to the Register, not to a website summary.

The eCFR is the place most practitioners read current regulations, and it is honest about its status. [The eCFR's own explainer](https://www.ecfr.gov/reader-aids/understanding-the-ecfr/what-is-the-ecfr) calls it "a web version of the Code of Federal Regulations (CFR) that we update daily" and "an editorial compilation of CFR material and amendments published in the daily Federal Register." It is maintained "as an informational resource," with the goal of eventually becoming an officially recognized publication. Until then, it says, "those relying on it for legal research should verify their results against the most current official edition of the CFR, the daily Federal Register, and the List of CFR Sections Affected (LSA)."

The U.S. Code has its own wrinkle. Under [1 U.S.C. § 204(a)](https://www.law.cornell.edu/uscode/text/1/204), the Code establishes "prima facie the laws of the United States," but titles enacted into "positive law" are "legal evidence" of the laws. In practice, if the Code's text for a non-positive-law title ever differs from the Statutes at Large, the session law controls, so a careful brief on a close textual point checks the enacted version.

Annotated codes are a separate product category. Commercial annotated codes add notes of decisions, cross-references, history and secondary-source pointers under each section, so you see how courts have applied a statute without running a separate search. Government sites publish the statutory text only. If your question is "what does the statute say," free text is enough; if it is "how have courts read this subsection," an annotated code saves hours.

### PACER and the PACER Case Locator

PACER is the federal judiciary's system for federal court dockets and filings, and the PACER Case Locator is, in the judiciary's words, "a nationwide index that searches all federal court case records in one central location." Fees are low per page but add up on searches and reports, which have no cap. The figures below come from the judiciary's [PACER pricing page](https://pacer.uscourts.gov/pacer-pricing-how-fees-work).

![PACER fee cheat sheet: 10 cents a page, a $3 cap per document, uncapped searches and transcripts, and free court opinions](/images/blog-content/legal-research-tools/04-pacer-fees-cheat-sheet.jpg "PACER fees as published by the federal judiciary, October 2026. The $3 cap is per document; searches and transcripts have no cap.")

| Item | Charge | Cap or exception |
|---|---|---|
| Documents | $0.10 per page | $3 per document for dockets, motions, orders, judgments and briefs |
| Searches | Per page generated, even "no matches found" | No maximum |
| Non-case-specific reports | Per page | No maximum |
| Transcripts | Per page | No maximum; added 90 days after production |
| Audio files | $2.40 per file | None stated |
| Quarterly bill | Only if over $30 | $30 or less in a quarter is waived |
| Court opinions | Free | Always free |
| Service Center record search | $30 per name or item | Per search |

Fees as published by the federal judiciary, verified October 22, 2026.

The $3 cap is per document, not per account or per session. The judiciary says search results are charged by pages generated "even if the search displays 'no matches found.' There is no maximum fee for these searches." There is also no maximum for non-case-specific reports or transcripts, and transcripts "are added to PACER 90 days after they are produced."

Several things are free. A party receiving a Notice of Electronic Filing gets one free copy of the filed document. Anyone can view case information at a federal courthouse at no charge. "Court opinions are always free," and courts may grant fee exemptions to "indigents, pro bono attorneys, academic researchers, and non-profit organizations." Users are billed quarterly only "if they accrue more than $30 in the previous quarter."

To keep costs down, search the RECAP Archive before you pay for anything. Run narrow Case Locator searches with a court, date range and party name, since broad searches generate uncapped pages. Save the free copy that comes with every NEF in your matter file. If your work qualifies, ask the court about an exemption.

## Free research access through your state bar

Every state bar gives members free access to vLex Fastcase or Decisis, according to [Duke Law's Goodson Law Library table](https://law.duke.edu/lib/statebarassociations) dated May 9, 2025. Ten states offer Decisis only, Indiana and Wyoming offer both, and the rest offer Fastcase. County and local bars may add more.

### Fastcase or Decisis: which states offer which

The Duke table says, "All U.S. state bar associations provide their members with free access to at least one of the two major low-cost legal research services, vLex Fastcase and Decisis," and adds that "county/local bar associations may also provide free access." Here is how the states split in that table.

| Bar benefit | States (Duke Law table, May 9, 2025) |
|---|---|
| Decisis only | Delaware, Kentucky, Maine, Maryland, Massachusetts, Nebraska, New Jersey, Ohio, Pennsylvania, Utah |
| Both Decisis and vLex Fastcase | Indiana, Wyoming |
| vLex Fastcase | All other states plus D.C. |

Source: Duke Law Goodson Law Library, as of May 9, 2025. Confirm on your bar's member-benefits page before relying on it.

Bar contracts change, so treat the table as a starting point and check your own bar's site. Casemaker is no longer a separate bar benefit: Fastcase says on its [Casemaker transition page](https://www.fastcase.com/casemaker/) that Casemaker users "are being upgraded to the award-winning Fastcase legal research platform," and no state in the Duke table lists Casemaker. [Fastcase says](https://www.fastcase.com/) it partners "with over 80 national, state, and county bar associations" as "a free member benefit." The bar version is a distinct product: vLex says its [Fastcase Library](https://vlex.com/fastcase-library) "is available exclusively through bar association partnerships," so a lawyer cannot buy that edition directly. Direct buyers look at vLex's paid products instead, covered in the paid-platform section above.

### How to activate and use bar-benefit access

Activation takes a few minutes, but the order of steps matters because the benefit is tied to your membership record.

1. Log in to your state bar's member portal with your bar credentials.
2. Find the legal research benefit link, usually under member benefits or practice resources.
3. Create your research account from that link. A direct sign-up on the vendor's site may not carry the bar benefit.
4. Confirm which jurisdictions, federal materials and secondary sources your bar's plan includes.
5. Set up saved searches and alerts for your practice areas and key courts.
6. Book the platform's training session if your bar offers one.

A few habits make a bar benefit go further. Run the citator on every case you plan to rely on; in Fastcase Library that is the Cert Citator, which vLex says "instantly reveals when cases have been overturned or criticized." Cross-check one key authority in a second source, such as CourtListener, to catch gaps. Before a deadline depends on it, note what your plan excludes, such as certain secondary treatises or out-of-state materials, so you are not discovering the gap the night before a filing.

### Free Westlaw and Lexis: trials, libraries and school accounts

There is no permanent free Westlaw or Lexis plan for practicing lawyers on the vendor pages read for this guide, as of October 2026. There are, however, several legitimate ways to get temporary or location-based access.

Vendor trials are the most direct path. LexisNexis advertises a [2-day free trial of Lexis+ with Protégé](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page). Thomson Reuters' [Westlaw Edge plans page](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge/plans-pricing?segment=lawfirm) lists "Free trials & demos" but does not state a trial length, so ask the sales rep how long yours will run before you schedule research around it. [Decisis](https://www.decisis.com/) offers "TRY BEFORE YOU BUY" and a trial option for lawyers outside its bar-benefit states. Confirm current trial terms with each vendor.

Some public, county and courthouse law libraries give walk-in patrons access to commercial databases on library terminals. Policies vary a great deal, so check the library's website or call the reference desk before you go, and ask whether printing or emailing results is allowed.

Law schools often provide academic accounts to students and, in some cases, limited alumni access. Ask your school's law library what it offers and on what terms.

If you want a low-cost paid option instead of a free one, [Descrybe's pricing page](https://descrybe.com/pricing) lists its Legal Engine at $25 a month and its Platform at $50 a month, as of October 2026. That is the lowest published paid AI research price on this page; confirm current pricing with the vendor.

## What free legal research costs you

Free sources cost time and risk rather than money. You give up a treatment citator, editorial headnotes and annotations, guaranteed currency, complete coverage and reliable alerts, so a free-only workflow needs extra verification steps before anything is filed.

### Five gaps in free sources

Each gap below has a workaround. The workarounds are real, but each adds steps.

| Gap | What you lose | Workaround |
|---|---|---|
| No treatment citator | You cannot see at a glance if a case was overruled; "Cited by" has no signals | Bar-benefit Cert Citator, Descrybe's Cytator, or one paid citator run before filing |
| No headnotes or annotations | Slower to find the exact point of law in a long opinion | Start in secondary sources, then read the case |
| Currency and completeness | Google disclaims completeness; the eCFR is not official | Verify against govinfo, the official CFR and the LSA |
| Coverage gaps | Scholar's state coverage starts in 1950; trial-court orders largely missing | PACER, state court portals, CourtListener |
| Limited alerts | Free alert caps on CourtListener | RECAP extension or a paid docket tracker |

The citator gap is the one that hurts most. Without treatment flags, the only way to know whether an opinion is still good law is to open the citing cases and read them. Descrybe names its citator the [Cytator](https://descrybe.com/), and the bar-benefit Cert Citator covers the same need for members, so most lawyers can close this gap without a premium subscription.

Headnotes are editorial aids, not law. Wex defines a [headnote](https://www.law.cornell.edu/wex/headnote) as a summary "written by third party publishers" that does "not form a part of the judgment" and is not binding precedent. Their value is speed: they point you to the right paragraph. Without them, a good treatise or practice guide does the same job, at the cost of an extra step.

The currency workaround costs the most time on regulatory questions. Checking an eCFR section against the LSA and the daily Federal Register is not hard, but it is a manual comparison you must remember to do every time. Coverage workarounds cost time in a different way: state court portals differ by state, and some charge per document. Alert workarounds are the cheapest, since the RECAP extension is free, but paid trackers become worth it once you follow more dockets than the free limits allow.

### When free is enough and when it is not

The deciding question is what happens if you miss something. When an overlooked authority would only cost you a second search, free sources are fine. When it could cost the client a claim or expose you to sanctions, add a citator-backed platform.

Free is usually enough for:

- reading a statute's or regulation's current text
- pulling a case you already have a citation for
- background reading in a new area of law
- federal opinions, which PACER and CourtListener provide free

Free is usually not enough for:

- briefs and dispositive motions
- multi-state surveys of deadlines or fault rules
- anything where an overruled case would hurt the client
- high-volume docket monitoring across many cases

Whatever tool you use, the professional standard does not change. Under [Federal Rule of Civil Procedure 11](https://www.law.cornell.edu/rules/frcp/rule_11), an attorney who presents a paper to the court certifies that its legal contentions are warranted after "an inquiry reasonable under the circumstances." A free tool does not lower that bar; it only changes how much of the checking you do by hand. The five-question test in the buying section below helps decide whether your practice has crossed the line where a paid platform pays for itself.

## Docket, analytics, verdict and expert research tools

Beyond case law, litigators buy specialist research: docket trackers for following cases and parties, litigation analytics on judges, courts and opposing counsel, verdict and settlement data for valuation, expert witness research, and investigative databases for locating people. Most are quote-only and sold separately from the core platform.

| Tool | Owner | What it does (vendor description) | Price model | Plaintiff use |
|---|---|---|---|---|
| PACER and Case Locator | Federal judiciary | Federal dockets and filings | Per page | Removal and MDL member cases |
| CourtListener and RECAP | Free Law Project | Free filings and alerts | Free | Following MDL dockets |
| CourtLink | LexisNexis | Federal and state dockets with alerts and tracking | Quote | A defendant's litigation history |
| Docket Alarm | vLex, part of Clio | Court records with alerts and analytics | Quote | Multi-court tracking |
| Lex Machina | LexisNexis | Federal and some state analytics on judges, parties, experts | Demo | Judge and expert history |
| Westlaw Edge analytics | Thomson Reuters | Judges, courts, attorneys, firms, damages | In Edge | Venue and damages |
| Docket Navigator | Centellic | Patent, IP, antitrust and securities | Free trial | Little for PI |
| Lexis Verdict & Settlement Analyzer | LexisNexis | Verdict and settlement patterns | Quote | Case valuation |
| Lexis Expert Research On-Demand | LexisNexis | Analyst research on experts, attorneys, judges | Quote | Expert vetting |
| Tracers | Tracers | Investigative research software | Quote | Locating witnesses |
| CLEAR | Thomson Reuters | Public-records investigations | Quote | Locating parties |

Vendor descriptions as published, October 22 and 24, 2026. Prices are quote-only unless stated.

### Docket research and alerts

Docket research tells you what is happening in a case right now: who filed what, which deadlines are set and which rulings came down. For federal courts, start with PACER and the Case Locator and use the RECAP Archive and its alerts to avoid paying twice (fees are covered in the PACER section above).

State courts are harder. Each state runs its own e-filing and case-search portal, with different search fields, fees and coverage of older cases. Commercial trackers aggregate both systems. LexisNexis lists [CourtLink](https://www.lexisnexis.com/en-us/products/courtlink.page) as "an AI-powered docket research and tracking service," and its [legal research hub](https://www.lexisnexis.com/en-us/solutions/legal-research.page) describes it as a "U.S. federal & state court docket database with alerts, tracking, profiling, search, and document retrieval." On [vlex.com](https://vlex.com/), vLex says Docket Alarm lets you "Search across 850+ million court records with real-time alerts" and "powerful APIs." Trellis and UniCourt are other examples of state-court docket and analytics services; ask each vendor which courts it covers in your states.

A workflow tip: set alerts on the parties and the judge, not only the case number. A defendant's new filings in other courts, or a judge's ruling in a related case, often matter before anything changes on your docket. Monitoring JPML and MDL dockets is its own discipline, covered in the plaintiff research stack below. For how docket filings turn into case criteria, see [qualification criteria drawn from MDL docket filings](/blogs/mass-tort-qualification-criteria).

### Litigation analytics on judges, courts and counsel

Litigation analytics turn past dockets into patterns. LexisNexis puts the distinction this way: "Litigation analytics analyzes historical case data and dockets to identify trends, judge behavior, settlement patterns, and outcome probabilities," as opposed to traditional research, which finds the law.

[Lex Machina](https://lexmachina.com/), owned by LexisNexis and "now empowered by Protégé," says it covers "all 94 federal district courts, 13 courts of appeal, PTAB, specialty venues, and various enhanced state courts," with state court motion metrics for "40+ motion types." It also promotes a Lex Machina 2026 Product Liability Litigation Report. Thomson Reuters says [Westlaw Edge](https://legal.thomsonreuters.com/en/westlaw) provides "data-driven insights on judges, courts, attorneys, law firms, damages, and case types." LexisNexis lists Context for "language analytics" on judges, courts, expert witnesses and companies.

Not every analytics tool fits injury work. [Docket Navigator](https://www.docketnavigator.com/) lists patent, copyright, trademark, antitrust, trade secret and securities litigation, with alerts that "include attorney-written summaries," and does not list product liability or personal injury.

In my view, a plaintiff firm gets the most from analytics in three places: choosing a venue when more than one is available, setting expectations for a judge's motion practice and timing, and learning how opposing counsel has settled similar cases. One caution: analytics describe past behavior, and small samples mislead. A judge with four product cases is an anecdote, not a trend.

### Verdict and settlement research for case value

Verdict and settlement research tells a contingency practice what similar cases have been worth. It supports case valuation at intake, demand letters, settlement authority discussions and, just as important, realistic client expectations.

LexisNexis lists Lexis Verdict & Settlement Analyzer as a tool that "visualizes verdict data to reveal settlement patterns, resolution trends & case strengths." Thomson Reuters includes damages among the Westlaw Edge analytics described above. Ask each vendor which jurisdictions and injury types its verdict data covers in depth, since reporting of settlements is voluntary and uneven.

Read verdict reports critically. A reported verdict is often the high point before remittitur, appeal or a confidential post-trial settlement, and many settlements are never reported at all. Match on the factors that move value in your cases: injury type and permanence, liability disputes, venue, insurance limits and the plaintiff's age and work history.

In mass torts, the same logic applies at scale: bellwether trial outcomes shape the negotiation of global settlements and the values assigned within them. For how that plays out, see [how bellwether verdicts and settlements set case values](/resources/mass-tort-settlement-guide).

### Expert witness research

Expert research checks whether an expert will hold up, yours or the other side's. The core questions are the same in every case: what the expert has testified to before, what they have published, how courts have treated their testimony, including any exclusions, and whether there is any licensing or disciplinary history.

LexisNexis offers Lexis Expert Research On-Demand, which it describes as "research analysts who vet, conduct, package, and deliver research on expert witnesses, attorneys, judges." That is an analyst service rather than a database you search yourself. Its Context product adds language analytics on expert witnesses. Lex Machina lists "6K+ expert witnesses" among its counts, according to its own site.

The scientific literature that supports general causation experts is a separate task, covered in the plaintiff research stack below.

### People and asset investigation databases

People-search and public-records databases are a different category from case-law research, and lists that mix them in with Westlaw and Lexis confuse buyers. They answer "where is this person and what do they own," not "what is the law."

[Tracers](https://www.tracers.com/) calls itself "Investigative Research Software" and lists law firms and legal professionals among its users, with use cases such as locating witnesses. Its footer says it "is not a 'Consumer Reporting Agency'" under the Fair Credit Reporting Act and that its services "may not be used as a factor in determining eligibility" for credit, insurance, employment or other FCRA purposes. Thomson Reuters describes [CLEAR](https://legal.thomsonreuters.com/en/products/clear-investigation-software), its investigation product, as "trusted public-records technology that reveals critical connections in one platform." Both are quote-only.

Plaintiff firms use these tools to locate witnesses, find defendants and the people an insurer covers, and check assets when collectability is in doubt. Federal limits apply, including the FCRA (15 U.S.C. § 1681b) and the Driver's Privacy Protection Act (18 U.S.C. § 2721); they are covered with claimant verification in the plaintiff research stack below.

AI assistants now sit on top of almost every tool in this guide, from citators to docket trackers, which raises the accuracy and ethics questions the next section takes up.

## AI legal research tools: accuracy, ethics and checks

AI research tools now sit inside Westlaw, Lexis, vLex and Descrybe, and they are useful for first drafts of research. They are not reliable enough to cite unchecked: a peer-reviewed study found leading tools hallucinated 17% to 33% of the time in their 2024 versions, and ABA Formal Opinion 512 treats independent verification as part of a lawyer's duty of competence.

![Six-step flowchart for checking AI legal research: pull citations, read sources, run a citator, check quotes, confirm, log](/images/blog-content/legal-research-tools/05-ai-citation-check-flow.jpg "Treat every AI research answer as a lead, not an authority. ABA Formal Opinion 512 ties verification to the duty of competence.")

This chapter covers how research-grade AI differs from a general chatbot, what the independent evidence on accuracy shows, the court decisions that sanctioned lawyers for fabricated citations, the ethics rules that apply, how to handle client files, and a six-step check to run on every AI answer before it reaches a brief. This is general information, not legal advice. States adopt the ABA Model Rules with their own changes, and courts and bars publish their own guidance, so confirm the rules that bind you in your jurisdiction and before each judge.

### Research-grounded AI versus general chatbots

The distinction that matters is where an answer comes from. A research-grounded tool retrieves documents from a defined legal database first and then writes an answer from what it retrieved, with links back to those documents. A general chatbot answers from patterns in its training data, and it can produce a case name, citation and quotation that look right and do not exist.

The large vendors sell the grounded model. Thomson Reuters says its Deep Research feature is ["grounded in Westlaw and Practical Law content"](https://www.thomsonreuters.com/en/press-releases/2025/august/thomson-reuters-launches-cocounsel-legal-transforming-legal-work-with-agentic-ai-and-deep-research) and that it "explains its process, sources its answers ... with human oversight." LexisNexis says [Protégé Legal AI](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page) "grounds all responses in LexisNexis sources and includes Shepard's citations." Both statements are vendor positioning, not independent findings.

Grounding is a property of a mode, not of a brand. The same LexisNexis page says "Protégé General AI includes access to large language models from OpenAI, Google, and Anthropic," which is a general-model mode inside a research platform. In the other direction, Descrybe says its [Legal Engine](https://descrybe.com/pricing) works inside ChatGPT, Claude, Microsoft 365 Copilot and Perplexity, and that its tools ["resolve citations, verify quotations, check subsequent treatment"](https://descrybe.com/). Before relying on any answer, know which mode produced it.

So is ChatGPT useful for legal research? For brainstorming issues, listing elements of a claim, suggesting search terms and outlining a memo, a general chatbot can save time. As a source of citations it is unsafe. The fabricated opinions in *Mata v. Avianca* came from ChatGPT, and Stanford HAI reported that an earlier study found general-purpose chatbots ["hallucinated between 58% and 82% of the time on legal queries"](https://hai.stanford.edu/news/ai-trial-legal-models-hallucinate-1-out-6-or-more-benchmarking-queries). For a wider look at where AI helps a firm, see [legal AI tools for drafting, intake and review](/blogs/legal-ai-tools) and [using Claude for legal work](/blogs/claude-for-lawyers).

There is a second reason plaintiff firms should understand these tools: injured people now research their own claims in the same chatbots, and those answers shape what prospects expect before they ever call. The guide to [how claimants research their claims in ChatGPT and AI Overviews](/blogs/aeo-for-personal-injury-lawyers) covers that side.

### What independent accuracy research found

The best-known independent test is by Magesh, Surani, Dahl, Suzgun, Manning and Ho, published as ["Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools"](https://arxiv.org/abs/2405.20362) in the *Journal of Empirical Legal Studies* (2025), with a preprint dated May 30, 2024. The authors describe it as "the first preregistered empirical evaluation of AI-driven legal research tools." They ran 202 queries and found that Lexis+ AI, Westlaw AI-Assisted Research and Ask Practical Law AI ["each hallucinate between 17% and 33% of the time"](https://dho.stanford.edu/wp-content/uploads/Legal_RAG_Hallucinations.pdf).

The same paper found that hallucinations "are reduced relative to general-purpose chatbots (GPT-4)." Grounding helped. It did not solve the problem.

The abstract also records the claims vendors made in 2023: "eliminating" hallucinations (Casetext), "avoid[ing]" them (Thomson Reuters) and "hallucination-free" citations (LexisNexis). The authors conclude that "the providers' claims are overstated."

Read the dates carefully. The 17% to 33% range measured 2024 versions of these products, before Lexis+ AI was renamed Lexis+ with Protégé and before Thomson Reuters launched CoCounsel Legal and Westlaw Advantage. As of October 2026 we found no independent, published benchmark of the current versions (Protégé, CoCounsel Legal's Deep Research or vLex's Vincent). Vendor accuracy numbers exist, but they are self-reported. Until an independent test appears, the safe assumption is that current tools are better than the 2024 versions in some ways and still wrong some of the time.

In practice, the reasoning is simple. If even a modest share of answers contains an error, an answer that cites eight or ten authorities is likely to contain at least one problem somewhere: a case that does not support the proposition, a misquoted line, a pin cite that points to the wrong page or a decision that has been reversed. That is why the check at the end of this chapter applies to every citation, not to a sample.

### The hallucinated-citation case record

Courts have now sanctioned lawyers in many jurisdictions for filing AI-generated citations they did not check. The four decisions below are a representative record drawn from two primary orders.

| Case | Court and date | What happened | Sanction |
|---|---|---|---|
| *Mata v. Avianca, Inc.* | S.D.N.Y., June 22, 2023 | Six fabricated opinions generated by ChatGPT; counsel stood by them | $5,000 jointly and severally, plus letters to the client and each falsely named judge |
| *Wadsworth v. Walmart Inc.* | D. Wyo., Feb. 24, 2025 | Motions in limine cited nine cases; eight did not exist; source was an in-house AI database | Pro hac vice revoked and $3,000 for the drafter; $1,000 each for two signers |
| *Hayes* | E.D. Cal., Jan. 17, 2025, as summarized in *Wadsworth* | AI-related citation errors | $1,500 |
| *Gauthier v. Goodyear Tire & Rubber Co.* | As summarized in *Wadsworth* | AI-related citation errors | $2,000 plus a generative-AI CLE |

In [*Mata v. Avianca, Inc.*](https://storage.courtlistener.com/recap/gov.uscourts.nysd.575368/gov.uscourts.nysd.575368.54.0_8.pdf), No. 22-cv-1461 (PKC), 678 F. Supp. 3d 443, Judge P. Kevin Castel found that counsel "submitted non-existent judicial opinions with fake quotes and citations created by the artificial intelligence tool ChatGPT, then continued to stand by the fake opinions." The court imposed "a penalty of $5,000 ... jointly and severally" and required letters to the client and to each judge falsely named as an author.

The record keeps growing. Damien Charlotin's independently maintained [AI Hallucination Cases Database](https://www.damiencharlotin.com/hallucinations/), which says it is updated daily, lists more than 2,000 decisions worldwide as of October 2026. Counts change daily, so check the database for the current figure.

*Wadsworth* carries the sharpest lessons for plaintiff firms. It was a personal injury case brought by plaintiffs. The fabricated cases came from the firm's own in-house AI database, not a public chatbot. And the [order by Judge Kelly H. Rankin](https://storage.courtlistener.com/recap/gov.uscourts.wyd.64014/gov.uscourts.wyd.64014.181.0_1.pdf) sanctioned the two attorneys who signed without reviewing the cited law, not only the drafter. The court held that signing without that review violates Rule 11, whose "reasonable inquiry" duty "remains the same" as technology changes. The same order describes *Hayes* and *Gauthier*, the source for the last two rows of the table.

Neither court condemned the technology itself. *Mata* says "there is nothing inherently improper about using a reliable artificial intelligence tool for assistance. But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings." The opinion opens by noting that "good lawyers appropriately obtain assistance from junior lawyers, law students, contract lawyers, legal encyclopedias and databases such as Westlaw and LexisNexis." AI is one more source of assistance, and the signer still owns the result.

### ABA Formal Opinion 512 and the duty to verify

The ABA Standing Committee on Ethics and Professional Responsibility issued [Formal Opinion 512, "Generative Artificial Intelligence Tools,"](https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf) on July 29, 2024. It lists legal research first among the tasks lawyers use generative AI for, and it notes that "for basic legal research, AI may enhance lawyers' search results." The opinion then walks through the duties AI use touches: competence, confidentiality, communication, supervision, meritorious claims, candor and reasonable fees.

On competence, the opinion says reliance on AI output "without an appropriate degree of independent verification or review" could violate Model Rule 1.1. It does not set one standard for every use: the level of verification "will necessarily depend on the GAI tool and the specific task." A brainstormed list of issues needs less checking than a citation in a summary judgment brief. The opinion cites the 17% to 33% finding from the 2024 study as part of its reasoning.

Competence also has a technology component. [Comment 8 to Model Rule 1.1](https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/comment_on_rule_1_1/) tells lawyers to keep abreast of "the benefits and risks associated with relevant technology." The ABA adopted that language in August 2012, and according to [LawSites' tracker](https://www.lawnext.com/tech-competence), "40 states, the District of Columbia and Puerto Rico" have adopted it in some form. It is a Model Rule comment, so it binds you only as your state has adopted it.

Candor is the second pillar. Opinion 512 states that "Rule 3.3 makes it clear that lawyers cannot knowingly make any false statement of law or fact to a tribunal," and it also discusses Rule 3.1 (meritorious claims) and Rule 8.4(c) (dishonesty and misrepresentation). A fabricated citation that reaches a court can implicate all three.

In federal court, [Fed. R. Civ. P. 11(b)](https://www.law.cornell.edu/rules/frcp/rule_11) adds a filing-level duty: by signing, an attorney certifies, after "an inquiry reasonable under the circumstances," that legal contentions "are warranted by existing law or by a nonfrivolous argument" for changing it. *Wadsworth* applied exactly that duty to AI-generated citations.

State variation matters here more than almost anywhere else in this guide. States adopt the Model Rules with changes, several bars and courts have published their own AI guidance, and some judges issue standing orders on AI use in filings. Check your state's rules and ethics opinions and each assigned judge's current requirements before you file. How to bill for AI-assisted research is a separate question, covered in the billing section later in this guide. This is general information, not legal advice.

### Confidentiality when you upload client files

Uploading a medical record, a deposition or a client's intake notes to an AI tool is a confidentiality decision, not only a research one. Opinion 512 says that for "self-learning" generative AI tools, which may use inputs to improve the model, "a client's informed consent is required prior to inputting information relating to the representation." It adds that adding "boiler-plate provisions to engagement letters ... is not sufficient."

That does not mean every AI use must be disclosed to every client. Where Model Rule 1.6 does not require informed consent, the opinion says the lawyer must "separately consider whether other Model Rules, particularly Model Rule 1.4, require disclosing the use of a GAI tool." That is a case-by-case judgment.

A working checklist before you upload anything:

- Read the vendor's terms on data retention, whether inputs are used for training and which subprocessors handle your data.
- Prefer modes the vendor documents as grounded and as not training on your inputs, and keep a copy of that documentation.
- Know where uploads go and how long they stay. LexisNexis says uploads of up to 10 documents that are not saved to a Vault ["are automatically purged at the end of each session"](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page).
- Keep medical records and other protected health information out of general consumer chatbots.
- Record client consent in the file where your analysis says it is required.
- Ask every vendor for its security certifications. vLex, for example, says its platform is ["SOC2 and ISO 27001 certified"](https://vlex.com/); ask for the current report, not just the claim.

### A six-step check for every AI research answer

This protocol matches the flowchart above. Run it on every authority an AI tool gives you before the authority goes into a memo, a demand letter or a filing.

1. Pull every cited authority (case, statute, rule) into a list.
2. Read the full source, not the AI summary or a headnote.
3. Run a citator on each authority (KeyCite, Shepard's, Cert Citator or Cytator).
4. Check every quotation and pin cite against the source text.
5. Confirm jurisdiction, date and procedural posture match your issue.
6. Log what you verified, when and in which tool, in the matter file.

Step 2 is where most errors are caught. AI summaries compress holdings, and a case that discusses your proposition in dicta, or rejects it, can read as support in a two-line summary. Step 5 catches a quieter error: a real, good-law case from the wrong jurisdiction or decided under a since-amended statute. In a multi-state docket, that mistake repeats across every claimant in the affected states.

Who does the check? A paralegal or associate can pull sources and run the citator, but the lawyer who signs owns the result, as *Wadsworth* shows. Supervision duties for non-lawyer staff are covered in the operations section later in this guide. The log in step 6 is what you show a court if a citation is ever challenged.

## The plaintiff firm research stack for PI and mass tort

Plaintiff firms research more than case law. A PI practice needs state deadline, fault and damages rules plus verdict data; a mass tort practice also tracks the JPML's MDL reports, transferee-court orders, FDA and CPSC adverse-event and recall data, product labels and the medical literature on causation, then turns all of it into qualification criteria that intake can apply.

![Mass tort research loop: JPML report, MDL dockets, adverse event and recall data, published science, criteria and intake](/images/blog-content/legal-research-tools/06-mass-tort-research-loop.jpg "Mass tort research runs on dockets, regulators and science as much as case law. The JPML listed 162 pending MDL dockets on 9/30/2026.")

This chapter maps each layer of that stack to the sources that answer it and the limits each source states about itself. It touches federal statutes on consumer data, so the usual caution applies: this is general information, not legal advice, and state law varies on almost every rule mentioned here.

### Everyday PI research: deadlines, fault rules and value

Most PI research questions repeat from file to file. Which limitation period applies, and does anything toll it (minority, incapacity, discovery, a claim against a government entity with its own notice deadline)? Does the state use pure comparative fault, modified comparative fault or contributory negligence? Is it a no-fault state, and has the client crossed the threshold to sue? Are there caps on non-economic damages, or on claims against particular defendants? What uninsured and underinsured motorist coverage applies, and which medical, Medicare, Medicaid or ERISA liens have to be resolved before money moves?

Each of those is a research task with a primary-law answer. The state's code and its appellate decisions on your core platform are the authority. A multi-state survey tool helps when you need to compare states (see the surveys section below), and verdict and settlement data help with value, as covered in the specialist tools section earlier in this guide.

For quick orientation, see the reference pages on [contributory negligence jurisdictions](/services/motor-vehicle-accident-leads/comparisons/contributory-negligence-states) and [no-fault vs at-fault state rules](/services/motor-vehicle-accident-leads/comparisons/no-fault-vs-at-fault). Treat any summary as a starting point. Re-verify the rule on the official code and the controlling cases before a filing deadline depends on it, because legislatures amend these statutes and courts reinterpret them.

### Mass tort docket monitoring: JPML, MDLs and alerts

Multidistrict litigation runs on one statute. Under [28 U.S.C. § 1407(a)](https://www.law.cornell.edu/uscode/text/28/1407), civil actions with "one or more common questions of fact" pending in different districts "may be transferred to any district for coordinated or consolidated pretrial proceedings," and each action "shall be remanded" at or before the end of pretrial proceedings. Section 1407(d) creates the Judicial Panel on Multidistrict Litigation: "seven circuit and district judges," "no two of whom shall be from the same circuit," with four needed to concur for action.

The numbers show why monitoring matters. The JPML's [pending MDL dockets report dated 9/30/2026](https://www.jpml.uscourts.gov/sites/jpml/files/Pending_MDL_Dockets_By_Circuit-September-30-2026.pdf) lists 162 MDL dockets with 204,967 actions pending, 717,515 total actions historically, 50 transferee districts and 141 transferee judges. Its [calendar-year 2025 statistics](https://www.jpml.uscourts.gov/sites/jpml/files/JPML_Calendar_Year_Statistics-CY-2025%201-16-26-Final_0.pdf) show products liability as the largest type at 61 of 158 pending MDLs (38%) as of Dec 31, 2025. In 2025 the panel held 6 hearings, decided 32 motions for centralization (21 granted, covering 278 actions; 16 denied, moot or withdrawn) and ordered 1,300 tag-along transfers.

What to monitor, and where:

1. The JPML pending-docket report and the [MDL Quarterly Caseload Summary](https://www.jpml.uscourts.gov/mdl-quarterly-caseload-summary), posted for March, June and September 2026, plus the panel's [statistics archive](https://www.jpml.uscourts.gov/statistics-info) back to 2006.
2. The [JPML's home page](https://www.jpml.uscourts.gov/) for hearing sessions, new motions for centralization and the current Panel Rules (effective February 19, 2026; Karen K. Caldwell is listed as Chair).
3. The transferee court's MDL page, where judges post case management orders and leadership appointments.
4. The MDL docket itself, through PACER or through [RECAP docket alerts](https://free.law/recap), which can also route ECF notices sent to an @recap.email address.
5. Bellwether scheduling orders, which set the pace of trials and often of settlement talks.
6. Plaintiff fact sheet and census orders, which define what each claimant must show and by when.

Pending and historical counts tell different stories, and reading them together is a basic skill. On the 9/30/2026 report, MDL 3029, *In re: Covidien Hernia Mesh Products Liability Litigation (No. II)* in the District of Massachusetts, shows 2,479 pending actions against 2,489 historical: nearly every case filed is still open. MDL 2753, *In re: Atrium Medical Corp. C-Qur Mesh Products Liability Litigation* in the District of New Hampshire, shows 101 pending against 3,592 historical: a large docket that has mostly resolved. A big historical number alone does not mean an active litigation. The overview of [hernia mesh litigation and its MDLs](/blogs/hernia-mesh-complications-lawsuit) shows how those two dockets sit side by side.

Mass Tort Marketing Agency publishes an [active mass tort litigation tracker](/mass-tort-leads); any status shown there, or on any other tracker, should be confirmed against the JPML report and the MDL court's own page before a firm acts on it.

### Product and regulatory databases for injury cases

Federal agencies publish the raw material for most product cases: adverse-event reports, device malfunction reports, product labels and consumer complaints. Each database also states what it cannot prove, and those caveats belong in your research notes.

| Database | Agency | What it holds | Update cadence | Built-in caveat |
|---|---|---|---|---|
| AEMS dashboard (formerly FAERS) | FDA | Drug adverse event reports from industry, providers and consumers; consolidating other product areas | Page current as of 03/11/2026; query tool described as coming | A report does not establish causation; no incidence rates |
| MAUDE | FDA | Medical device reports, last ten years | Monthly | Not for rates; does not show the device caused the event |
| DailyMed | NLM | Current labeling submitted to FDA, including boxed warnings and adverse reactions | Most recent submitted label | Shows today's label, not past versions |
| SaferProducts.gov | CPSC | Public reports of harm from consumer products | Maker notified within 5 business days; posted by the 10th business day after | CPSC does not guarantee accuracy |
| NHTSA complaints and recalls | NHTSA | Vehicle and equipment complaints and recalls | Not verified here | Check the agency's own notes |

The [FDA Adverse Event Monitoring System (AEMS) Public Dashboard](https://www.fda.gov/drugs/fda-adverse-event-monitoring-system-aems/fda-adverse-event-monitoring-system-aems-public-dashboard), formerly the FAERS dashboard, collects reports from "the pharmaceutical industry, healthcare providers and consumers." FDA's own page says "Existence of a report does not establish causation" and that the reports cannot estimate how often events occur. The [MAUDE database](https://www.fda.gov/medical-devices/mandatory-reporting-requirements-manufacturers-importers-and-device-user-facilities/about-manufacturer-and-user-facility-device-experience-maude-database) holds "the last ten years of MDR data," updated "every month," with each report labeled Manufacturer, Importer, User Facility or Voluntary (voluntary reports carry an "MW" prefix). Some narrative text is redacted under FOIA exemptions (b)(4) and (b)(6). [DailyMed](https://dailymed.nlm.nih.gov/dailymed/about-dailymed.cfm), run by the National Library of Medicine, carries the "most recent submitted labeling," including "boxed warnings, indications, … warnings and precautions, adverse reactions."

For consumer products, [SaferProducts.gov](https://www.saferproducts.gov/About) is "the CPSC-owned website where the public can file and read safety-related complaints about consumer products," created under the CPSIA of 2008. Under [15 U.S.C. § 2055a](https://www.law.cornell.edu/uscode/text/15/2055a), CPSC sends each report to the manufacturer "not later than 5 business days" after receipt and posts it "not later than the 10th business day" after transmittal.

Plaintiff firms use these sources for four jobs: spotting a signal (a cluster of similar reports), identifying the exact product and model, building a timeline of label changes and warnings, and corroborating a client's account of what happened. Two case write-ups show [how MAUDE adverse-event reports feed a device case](/blogs/bard-powerport-lawsuit) and how [AEMS (formerly FAERS) reports in the GLP-1 litigation](/blogs/ozempic-lawsuit-gastroparesis) were read.

The limits are stated by the agencies themselves. FDA says an AEMS report does not establish causation and cannot measure frequency, MAUDE data are "not intended to be used either to evaluate rates of adverse events" and a device report "does not necessarily demonstrate that the device caused" the event, and CPSC "does not guarantee the accuracy, completeness, or adequacy" of SaferProducts.gov content. Reports start research. They do not prove a case.

### Scientific literature for general causation

Every drug and device case has two causation questions. General causation asks whether the product can cause the injury at all, in anyone. Specific causation asks whether it caused this injury in this claimant. General causation is decided largely on published science, so mass tort research includes a literature search.

[PubMed](https://pubmed.ncbi.nlm.nih.gov/about/) is "a free resource" with "more than 40 million citations and abstracts of biomedical literature." It does not include full-text journal articles, so a PubMed result is a pointer to the paper you still have to obtain. ClinicalTrials.gov is the federal registry for clinical studies and is the place to check whether trials of a product were registered, completed or stopped. For expert discovery and testimony history, the specialist tools covered earlier in this guide, such as [Lexis Expert Research On-Demand and Context](https://www.lexisnexis.com/en-us/solutions/legal-research.page), are the next layer.

Three habits pay off. Search for systematic reviews, meta-analyses and epidemiological studies first, because case reports alone carry little weight. Record the publication date of each key study, since the timing of public knowledge often matters to discovery-rule and notice arguments. And save the full-text PDF with the matter, not just the link.

### Multi-state surveys for claimant inventories

A mass tort docket, or a multi-state PI practice, needs 50-state answers. Limitation periods, discovery rules and tolling vary by state, so a claimant inventory has to be sorted by state of residence or injury before anyone can say which claims are timely.

Two research platforms market tools for this. Thomson Reuters lists AI Jurisdictional Surveys as part of [Westlaw Edge with AI-Assisted Research](https://legal.thomsonreuters.com/en/westlaw), and its [plans page](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge-aar/plans-pricing?segment=lawfirm) says "All plans include AI-Assisted Research and AI Jurisdictional Surveys." vLex says Vincent can "Generate 50-state surveys instantly." Both are vendor descriptions.

Treat any AI-generated survey as a draft table. Verify each state's answer on the official code and the controlling cases, using the six-step check above, before it decides which claimants a firm signs or declines. For motor vehicle claims, the [state-by-state MVA qualification index](/tools/state-qualification-index) is a quick cross-reference, not a substitute for the statute.

### Claimant verification and people-search limits

Verification research confirms that a claimant is who they say they are, can be reached, actually used or was exposed to the product, and has treatment records that can be obtained with a signed authorization. Most of that work runs through intake documents and records requests. Some firms also use people-search and public-records tools, covered in the specialist tools section earlier in this guide, to locate claimants and witnesses.

Those tools carry legal limits that a research team should know. Under the Fair Credit Reporting Act, [15 U.S.C. § 1681b(a)](https://www.law.cornell.edu/uscode/text/15/1681b), a consumer reporting agency "may furnish a consumer report under the following circumstances and no other." [Tracers](https://www.tracers.com/), for example, says it "is not a 'Consumer Reporting Agency'" under the FCRA and that its services "may not be used as a factor in determining eligibility for credit, insurance, employment or another purpose in which a consumer report may be used."

Motor vehicle records have their own statute. The Driver's Privacy Protection Act, [18 U.S.C. § 2721](https://www.law.cornell.edu/uscode/text/18/2721), generally bars knowing disclosure of DMV personal information, but subsection (b)(4) permits use "in connection with any civil, criminal, administrative, or arbitral proceeding … including the service of process, investigation in anticipation of litigation, and the execution or enforcement of judgments and orders."

The research-hygiene answer is procedural. Write a permissible-purpose policy for each tool, log the purpose of every search, keep searches tied to a matter, and have counsel sign off on how the firm uses consumer data. Whether a particular use complies is a legal question for your counsel, not a research shortcut.

### A research scorecard for vetting a new tort

Before a firm spends on advertising or case acquisition for a new tort, its research should answer eight questions. The scorecard below puts each one next to the source that answers it and what a stronger or weaker answer looks like.

| Factor | Where to research it | Stronger signal | Weaker signal |
|---|---|---|---|
| Litigation stage | JPML report, panel orders | Centralized MDL with tag-alongs | No MDL or centralization denied |
| Docket trend | JPML pending counts over time | Pending actions rising | Pending far below historical |
| Trial calendar | Transferee court orders | Bellwether dates set | None scheduled |
| Regulatory action | FDA, CPSC, NHTSA records; DailyMed label history | Label change or recall | Reports only, no agency action |
| Science | PubMed, published epidemiology | Consistent peer-reviewed studies | Case reports only |
| Claim deadlines | State codes and surveys | Most claims timely in key states | Many claims time-barred |
| Proof of use and injury | Plaintiff fact sheet orders | Records claimants can obtain | Proof hard to document |
| Advertising rules | Model Rules 7.1 to 7.3, state rules, TCPA | Clear path | State-specific restrictions |

A framework for research questions, not a prediction of case value or outcome.

Score each factor from the primary sources, write down the date and the source behind each score, and revisit the whole sheet monthly while the docket moves. A tort can move from weak to strong in one panel order, or the reverse in one Daubert ruling. Two factors deserve their own deeper work. Deadlines and proof requirements are covered in [the research to do before launching a new tort](/resources/mass-tort-launch-checklist). Advertising rules under ABA Model Rules 7.1 to 7.3, state bar rules and the TCPA are covered in [researching bar advertising and TCPA rules](/mass-tort-marketing-compliance).

### Turning research into qualification criteria and intake

Research only helps the firm when it becomes criteria intake can apply. Each criterion should trace to a source: the product and exposure window (from the complaint, the MDL orders and the label history), the qualifying diagnosis (from the plaintiff fact sheet order and the science), the timing between use and injury, the documents that prove each element, and a deadline screen by state.

A simple template keeps every criterion tied to its evidence:

- Criterion: the element the claimant must meet, such as use of a named product within a date range.
- Source: the order, label or study that sets it.
- Intake question: the plain-language question a screener asks.
- Proof document: the record that confirms the answer, such as pharmacy records or an operative report.
- Disqualifier: the answer that ends the screen, such as a diagnosis outside the window.

For worked examples, see [qualification criteria by litigation, sourced from MDL docket filings](/blogs/mass-tort-qualification-criteria), and for the call-center side, [turning qualification research into intake questions](/blogs/mass-tort-intake-guide). Mass Tort Marketing Agency's [claimant verification service](/services/mass-tort-lead-verification) screens claimants against published qualification criteria, so the quality of the research upstream decides the quality of the leads downstream.

With the research layers mapped, the remaining question is which platform and stack to buy.

## How to choose a legal research platform

Choose in this order: confirm what your bar already gives you for free, define the jurisdictions and content you need, decide whether you need a paid citator and an AI layer, trial the finalists on real questions, and negotiate the contract before the trial ends. Firm size and practice decide the stack.

![Research stacks by firm type: solo PI, small PI firm, mass tort firm and large litigation firm, from free sources to AI](/images/blog-content/legal-research-tools/07-research-stacks-by-firm.jpg "Suggested research stacks by firm type. Treat them as starting points for trials, not vendor endorsements.")

Most buying mistakes come from skipping the first two steps. A lawyer who never activated the bar benefit compares a paid platform against nothing, so every demo looks like a bargain. A firm that never wrote down its jurisdictions buys all-states coverage it does not use, or a single-circuit plan that leaves out the state where half its cases are filed. The sections below give you a short test for whether paying is justified at all, a trial checklist, the contract terms worth negotiating, and suggested stacks by firm type. Treat the stacks as starting points for a trial, not as endorsements.

### A five-question test for whether you need a paid platform

Answer each question yes or no for the way your firm actually practices today, not the way you hope to practice next year.

1. Do you file briefs or dispositive motions where an overruled case would hurt the client? If yes, a full citator matters. Thomson Reuters' help page says a red KeyCite flag [warns that a case is no longer good law](https://www.thomsonreuters.com/en-us/help/drafting-assistant/westcheck/keycite-status-flags) for at least one point of law, and that is the warning you are paying for.
2. Do you practice in more than one state or federal circuit? If yes, coverage tiers drive price. Westlaw's [Edge with AI-Assisted Research plans](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge-aar/plans-pricing?segment=lawfirm) price single-circuit and all-states coverage differently.
3. Do you need secondary sources or practice guides every week? Treatises and practice guides are where paid platforms differ most from free sites.
4. Do you track dockets, judges or verdicts to value cases? If yes, you need the docket, analytics and verdict tools described earlier, which are usually separate products.
5. Will lawyers use AI research tools, and can you verify what they produce? If yes, you need a grounded AI layer plus the six-step check from the AI section, not a general chatbot.

My scoring rule: two or more "yes" answers usually justify trialing a paid platform. Zero or one usually means the bar benefit plus free sources is enough, with a paid citator check, for example at a law library, on the cases that carry a brief. Firms whose "yes" on question 4 comes from mass torts should pair this test with [a pre-launch tort research checklist](/resources/mass-tort-launch-checklist), because the research load for a new tort is mostly dockets, regulatory data and science, not case law.

### Features to test during a free trial

Write the checklist before the trial starts, then score every platform against the same twelve items using the same research questions. That is the fair side-by-side method from the Westlaw vs LexisNexis section, applied to every finalist.

- **Natural-language and terms-and-connectors search.** Good looks like both modes returning the controlling case on a question you already know the answer to.
- **Jurisdiction, date and court filters.** You should be able to restrict to your state's appellate courts and a date range in two or three clicks.
- **Citator signals and negative-treatment detail.** A flag alone is not enough; you want to open the citing case and read the sentence that caused the flag.
- **Headnotes and links to secondary sources.** Each point of law should lead to related cases and to commentary that explains it.
- **Saved searches and alerts.** Check that an alert can run on a search and on a single authority, and that it emails the right person.
- **Research history and folders.** History should survive logouts and be searchable weeks later, and folders should be shareable inside the firm.
- **PDF and Word export with citations.** Exports should keep the citation, court and date on every page.
- **Word add-in or brief checking.** Upload a filed brief and see whether the tool catches a citation you know has negative treatment.
- **AI answers with linked sources.** Every proposition should link to a source you can open, and the answer should say when it found nothing.
- **Document upload and analysis.** Test with a non-confidential document first, and read the data terms before uploading anything from a client file.
- **Integration with your practice management system.** Research should save to the matter, not to one lawyer's personal folder.
- **Admin controls.** Look for single sign-on, seat management and usage reports that show who used what, so you can tell at renewal whether every seat earned its cost.

### Pricing models and contract terms to negotiate

Research platforms use six pricing models, and each one shifts risk differently. Every price below is as of October 2026; confirm it with the vendor before you rely on it.

- **Published monthly plans with term discounts.** Westlaw lists starting prices online, but [only for new customers at firms with up to 10 attorneys](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge/plans-pricing?segment=lawfirm); larger firms contact sales. Edge with AI-Assisted Research starts at $155.35 a month for one circuit or $266.50 for all states and federal on a 3-year term, and [Westlaw Advantage](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-advantage/plans-pricing?segment=lawfirm) starts at $256.75 or $399.75. Thomson Reuters says 2-year plans carry a 12% discount and 3-year plans 18%.
- **Customized quotes.** LexisNexis says [Lexis+ with Protégé pricing is customized](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page); Bloomberg Law and vLex are also sold through a demo or quote.
- **Online-store pricing for small firms.** LexisNexis says [1 to 3 attorney firms can view costs and buy online](https://www.lexisnexis.com/en-us/small-law-firms.page).
- **Bar benefits.** Fastcase or Decisis, free with membership.
- **Flat monthly AI tools.** [Descrybe's Legal Engine is $25 a month](https://descrybe.com/pricing), renewing monthly, and its page says payments are non-refundable.
- **Annual per-user contracts.** [Law Insider lists $1,790 per user per year](https://www.lawinsider.com/pricing) after a 7-day trial.

Before you sign, work through this contract negotiation checklist:

- Term length and the exact discount it buys, in writing.
- The auto-renewal notice window, entered in the firm calendar on signing day.
- Annual price escalators, capped or fixed for the full term.
- Out-of-plan charges, and how the platform warns you before you open content your plan excludes.
- Jurisdiction scope, including what "all states" actually contains.
- Seat count, seat swaps and what happens when a lawyer leaves.
- Which AI features are included and which are add-ons.
- Data handling and model-training terms for uploaded documents.
- Early termination rights and what happens on a firm merger.
- Bar-member or association discounts.

A simple break-even test helps. Divide the monthly price by the value of one hour of lawyer time at your firm; the result is the number of hours a month the platform must save to pay for itself. Run it with your own figures for each finalist.

### Research stacks by firm size and practice

The table pairs a core platform with free, docket and AI layers for five common plaintiff and litigation practices. It summarizes the earlier sections in one place.

| Firm type | Core platform | Free layer | Dockets and analytics | AI research layer |
|---|---|---|---|---|
| Solo PI | Bar-benefit Fastcase or Decisis | Google Scholar, CourtListener, state codes | PACER as needed | Descrybe or none |
| Small PI firm, 2 to 10 lawyers | Bar benefit plus a single-circuit paid plan | Free government sources | RECAP alerts, verdict data as needed | Platform AI with the six-step check |
| Mass tort firm | Enterprise platform, all-states coverage, bar benefit as a backup | JPML reports, FDA and CPSC databases, PubMed | CourtLink or Docket Alarm, analytics | AI surveys verified state by state |
| Large litigation firm | Two enterprise platforms or one plus analytics | Govinfo for authenticated text | Full docket and analytics suite | Grounded AI under a written policy |
| Workers' comp or med mal practice | Bar benefit or single-state paid plan | State agency and board sites | State court portals | Optional |

Starting points for trials, not endorsements. Confirm content scope and price with each vendor.

The solo row leans on the bar benefit because Duke Law's Goodson Law Library reports that [all U.S. state bar associations give members free access](https://law.duke.edu/lib/statebarassociations) to vLex Fastcase or Decisis. The mass tort row is heavier because criteria shift as court orders come down. That holds even for firms that outsource claimant screening to [mass tort intake teams](/mass-tort-intake): someone inside the firm still has to read the docket, the regulator's data and the science, and update the questions intake asks.

## Running research inside the firm: workflow and billing

Research only pays off when it is saved where the next lawyer can find it, supervised when nonlawyers or AI tools do it, and billed in line with the ethics rules. This is general information, not legal advice; states adopt the ABA Model Rules with variations.

### Saving research to the matter and building a brief bank

Store research by matter first and issue second, so a folder path reads like "Client v. Defendant / Statute of limitations / Tolling." Name saved searches by issue and jurisdiction rather than by date or lawyer, because the next person will search for the issue. A brief bank holds filed briefs, memos and orders, tagged by court, judge, issue and filing date, with the outcome noted when it is known.

One rule matters more than any tag: re-run the citator on every authority before reusing it from the bank. A memo written two years ago can cite a case that has since drawn negative treatment. Track a few measures to see whether the system works: time to first memo on a new issue, citator rework found at review, and how often the same issue is researched twice. Adopt your own baselines; there is no published benchmark worth copying.

### Supervising paralegals, clerks and AI tools

ABA Model Rule 5.3(a) requires partners and managers to [make reasonable efforts to ensure the firm has measures](https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_5_3_responsibilities_regarding_nonlawyer_assistant/) giving reasonable assurance that a nonlawyer's conduct fits the lawyer's professional obligations. That covers paralegals and clerks doing research. [ABA Formal Opinion 512](https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf) lists supervision among the duties lawyers must consider when using generative AI. A short written research policy turns both into practice:

- Which research tools are approved, and for which tasks.
- Which tools or modes may receive client data, following the confidentiality section above.
- Who signs off on authorities before a filing goes out.
- A log of the six-step check for every AI-assisted research answer used in work product.
- Training on the citator and the AI tools for every new hire, recorded with a date.

Your state's version of Rule 5.3 and any local guidance on AI may add requirements, so check them before adopting the policy.

### Billing research costs to clients

The ABA baseline treats ordinary research subscriptions as overhead. ABA Formal Opinion 93-379, which the [Alabama State Bar adopted as stating its rule](https://www.alabar.org/office-of-general-counsel/formal-opinions/1994-02/), says library and similar costs are "subsumed within" the lawyer's charges unless the client was told otherwise in advance. ABA 512 repeats that lawyers may not bill general overhead, including the routine costs of "maintaining a library."

ABA 512 adds AI-specific points. Hourly billers must bill their actual time. A lawyer may not charge a client to learn an AI tool the lawyer will use regularly. And charging the same flat or contingent fee with and without the tool may be unreasonable under Rule 1.5 when AI saves substantial time.

For dated context only, Legal Tech Monitor's [January 2020 summary of the ABA Legal Technology Survey Report](https://www.legaltechmonitor.com/2020/01/aba-report-2020-legal-research-trends/) said 60% of lawyers did not bill research separately or folded it into hourly rates, while 25% billed it. In contingency practice, read the fee agreement's cost provisions and your state's rules before treating any database charge as a case cost.

## Sources and methodology

This guide was researched from primary sources between October 22 and 24, 2026: vendor product, pricing and help pages, court and agency records, statutes and rules, ABA ethics opinions and a peer-reviewed study. Nothing was hands-on tested.

Method. The search results for "legal research platforms", "legal research tools", "online legal research platform" and "legal research software" were reviewed on Google US on October 22, 2026, and competing guides were read for coverage. Every vendor price and feature was read on the vendor's own page that day. Ownership changes come from press releases. Secondary sources (LawSites, Duke Law's library table and Legal Tech Monitor) were used only where a primary page was unavailable, and they are attributed in the text.

Limits. Some prices load by script (base Westlaw Edge, the Lexis online store, CoCounsel and Practical Law), and some products are quote-only, so no price appears for them here. Several vendor sites blocked automated access; their claims are left out rather than guessed. The AI accuracy study measured 2024 product versions, not the products sold today.

Disclosure. The author is Tarun, founder of Mass Tort Marketing Agency, a personal injury and mass tort marketing agency. The agency has no affiliate, referral or sponsorship arrangement with any vendor named, and the picks are opinion based on documented features. More [about the agency](/about). Prices and product names are reviewed quarterly, and the JPML counts are updated when each new report posts.

Key primary sources:

- [Westlaw product page](https://legal.thomsonreuters.com/en/westlaw) and plans and pricing pages (read October 22, 2026)
- [KeyCite product page](https://legal.thomsonreuters.com/en/products/westlaw/keycite) and KeyCite status flags help page (October 22, 2026)
- Lexis+ with Protégé product page and small-firm page (October 22, 2026)
- [Clio's release on completing the vLex acquisition](https://www.clio.com/about/press/clio-completes-landmark-1b-vlex-acquisition-series-g-5b-valuation/) (November 10, 2025)
- [vLex Library](https://vlex.com/vlex-library) and [Fastcase Library](https://vlex.com/fastcase-library) pages (October 24, 2026)
- Duke Law's state bar research-access table (dated May 9, 2025)
- [PACER fee schedule](https://pacer.uscourts.gov/pacer-pricing-how-fees-work) (October 22, 2026)
- [JPML pending MDL dockets by circuit report](https://www.jpml.uscourts.gov/sites/jpml/files/Pending_MDL_Dockets_By_Circuit-September-30-2026.pdf) (September 30, 2026)
- [FDA AEMS public dashboard](https://www.fda.gov/drugs/fda-adverse-event-monitoring-system-aems/fda-adverse-event-monitoring-system-aems-public-dashboard) and [MAUDE database page](https://www.fda.gov/medical-devices/mandatory-reporting-requirements-manufacturers-importers-and-device-user-facilities/about-manufacturer-and-user-facility-device-experience-maude-database)
- [Magesh et al., hallucination study](https://arxiv.org/abs/2405.20362), *Journal of Empirical Legal Studies* (2025)
- [*Mata v. Avianca* sanctions order](https://storage.courtlistener.com/recap/gov.uscourts.nysd.575368/gov.uscourts.nysd.575368.54.0_8.pdf) (June 22, 2023) and the *Wadsworth v. Walmart* order (February 24, 2025)
- ABA Formal Opinion 512 (July 29, 2024)

## Frequently asked questions

Short answers to the questions lawyers ask most about research platforms; each one is covered in more depth in the sections above.

### What is the best legal research platform?

There is no single best platform for every firm. For full-service research, the two default choices are Westlaw, which Thomson Reuters sells in [three versions](https://legal.thomsonreuters.com/en/westlaw), and [Lexis+ with Protégé](https://www.lexisnexis.com/en-us/products/lexis-plus-ai.page), which pairs the Protégé AI assistant with Shepard's. For a tight budget, the free bar benefit is the starting point: Duke Law's library table reports that every state bar gives members [vLex Fastcase or Decisis](https://law.duke.edu/lib/statebarassociations). The right pick depends on your jurisdictions, your practice area, how often you file briefs and what you can spend. Trial the finalists on the same real questions before you decide.

### What is the best website for legal research?

It depends on what you are looking for. For free case law, [Google Scholar](https://scholar.google.com/intl/en/scholar/help.html) covers state appellate and supreme court opinions since 1950 and federal courts since 1923, and [CourtListener](https://www.courtlistener.com/help/coverage/) adds dockets and filings from the RECAP Archive. For statutes and plain-English definitions, Cornell's Legal Information Institute and its [Wex dictionary](https://www.law.cornell.edu/wex) are the usual stops. For federal text you can cite with confidence, use [govinfo's authenticated PDFs](https://www.govinfo.gov/about/authentication). None of these free sites includes a treatment citator.

### What are some good free legal research sites?

The most useful free sites each come with one caveat. Google Scholar is broad, but Google says it "does not warrant" completeness or accuracy. CourtListener and [RECAP](https://free.law/recap) offer free dockets and alerts up to a limit. [Case.law](https://lil.law.harvard.edu/blog/2024/03/26/transitions-for-the-caselaw-access-project/) hosts Caselaw Access Project bulk data, but not search. Cornell LII is a strong secondary source but not an official text. Govinfo's signed PDFs are authenticated. The [eCFR](https://www.ecfr.gov/reader-aids/understanding-the-ecfr/what-is-the-ecfr) is updated daily but is an editorial compilation, not yet official. PACER charges for most documents, though court opinions are always free.

### Can I use Westlaw for free?

Not permanently. Thomson Reuters' pages show no free plan, though its Edge plans page lists "Free trials & demos" without stating a trial length. Law school accounts, law libraries and courthouse libraries are other routes, with access rules that vary by library. State bars give members Fastcase or Decisis, not Westlaw. If you buy, Westlaw's published plans start at [$155.35 a month](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-edge-aar/plans-pricing?segment=lawfirm) for single-circuit Edge with AI-Assisted Research on a 3-year term, as of October 2026, and only for new customers at firms with up to 10 attorneys. Confirm with the vendor.

### What is the cheapest legal research platform?

The cheapest is the one you already pay for through bar dues. Duke Law's table reports every state bar offers Fastcase or Decisis free, with 10 states offering Decisis only as of May 2025. Among paid AI research tools, the lowest published price found for this guide is [Descrybe's Legal Engine at $25 a month](https://descrybe.com/pricing), as of October 2026; confirm with the vendor. Free government sources such as govinfo and court websites cost nothing, but they lack a citator, so you still need a way to check whether a case remains good law before you cite it.

### What do lawyers use for legal research?

Westlaw and Lexis remain the default paid platforms. The *Mata v. Avianca* sanctions order even opens by describing [databases such as Westlaw and LexisNexis](https://storage.courtlistener.com/recap/gov.uscourts.nysd.575368/gov.uscourts.nysd.575368.54.0_8.pdf) as ordinary help for good lawyers. The best usage data available is old: Legal Tech Monitor's [January 2020 summary of the ABA survey](https://www.legaltechmonitor.com/2020/01/aba-report-2020-legal-research-trends/) found Westlaw and Westlaw Edge preferred by 49% of respondents and Lexis Advance by 28% among fee-based tools, and 65% frequently used free tools. Treat those as history. Many solo and small-firm lawyers also rely on bar-benefit Fastcase or Decisis.

### Is Westlaw or LexisNexis better?

Neither is better for every firm. They differ in citators, AI layers, add-ons and how small firms buy. Westlaw's KeyCite includes [Overruling Risk](https://legal.thomsonreuters.com/en/products/westlaw/keycite), which Thomson Reuters says flags points of law implicitly undermined by an overruled decision. A LexisNexis guide describes [Shepard's signals for seven treatments](https://www.lexisnexis.com/supp/lawschool/resources/be-confident-shepards.pdf). Westlaw publishes online prices for new firms of up to 10 attorneys, while LexisNexis lets [1 to 3 attorney firms buy online](https://www.lexisnexis.com/en-us/small-law-firms.page). Run both trials on the same questions and score them on the same checklist.

### Is ChatGPT good for legal research?

It is useful for brainstorming search terms and issues, but unsafe as a source of citations. In *Mata v. Avianca*, a 2023 federal case in New York, counsel was sanctioned after filing a brief with fake cases produced by ChatGPT. Stanford HAI's write-up [cites an earlier study](https://hai.stanford.edu/news/ai-trial-legal-models-hallucinate-1-out-6-or-more-benchmarking-queries) in which general-purpose chatbots hallucinated 58% to 82% of the time on legal queries. If you use a chatbot at all, treat every case name and quotation as unverified until you have pulled the opinion and checked it in a citator.

### What are some popular legal research platforms?

The best-known paid platforms are Westlaw, Lexis+ with Protégé and [Bloomberg Law](https://pro.bloomberglaw.com/), which says it puts news, primary and secondary sources and practice tools "all in one place, for one price." Others include [vLex](https://vlex.com/), now owned by Clio, with Vincent AI, Docket Alarm and Fastcase; [Decisis](https://www.decisis.com/), which offers a trial; and Descrybe for low-cost AI research. On the free side, Google Scholar and CourtListener are the most common. Popularity is a weak test; check coverage for your jurisdictions instead.

### How much does Westlaw cost per month?

As of October 2026, Thomson Reuters publishes starting prices only for new customers at firms with up to 10 attorneys. Edge with AI-Assisted Research starts at $155.35 a month for single-circuit coverage and $266.50 for all states and federal, both on a 3-year term. [Westlaw Advantage](https://sales.legalsolutions.thomsonreuters.com/en-us/products/westlaw-advantage/plans-pricing?segment=lawfirm) starts at $256.75 and $399.75. Thomson Reuters says 2-year plans carry a 12% discount and 3-year plans 18%. The base Edge price loads by script and was not captured. Firms with more than 10 attorneys get a quote. Confirm with the vendor.

### How much does Lexis+ with Protégé cost?

LexisNexis does not publish a single price. It says Lexis+ with Protégé pricing is customized and depends on organization size, capabilities and content scope. Firms of 1 to 3 attorneys can view costs and buy from the LexisNexis Online Store, where orders can take up to 72 hours to process. LexisNexis also offers a 2-day free trial. No figure appeared in the text of the pages read for this guide in October 2026, so ask for a written quote and confirm it with the vendor before you compare it with Westlaw's published plans.

### Is Fastcase free for lawyers?

Often, yes, through your bar. Fastcase says it partners with [over 80 bar associations](https://www.fastcase.com/) to provide research as a free member benefit, and vLex says [Fastcase Library is available only through bar partnerships](https://vlex.com/fastcase-library), so you cannot subscribe to it directly. Not every state offers it: Duke Law's table, dated May 2025, lists 10 states where the bar benefit is Decisis only and 2 that offer both. Check your own bar's member-benefits page before assuming which service you have.

### What happened to Casetext?

Thomson Reuters [bought Casetext on August 17, 2023](https://www.thomsonreuters.com/en/press-releases/2023/august/thomson-reuters-completes-acquisition-of-casetext-inc.html) for $650 million in cash. According to a migration notice Thomson Reuters has since removed, legacy Casetext users moved to CoCounsel 2.0 in the first quarter of 2025. Today casetext.com redirects to Thomson Reuters' [CoCounsel page](https://www.thomsonreuters.com/en/cocounsel), and no standalone Casetext research product is sold. Any guide still quoting Casetext subscription prices is out of date.

### Is the Caselaw Access Project still searchable?

Not on its own site. Harvard's Library Innovation Lab wound down the Caselaw Access Project's search function and API in 2024 and pointed users to CourtListener, which now includes all CAP cases in its search engine. The lab still hosts bulk data and readable cases at case.law, which suits researchers who want to download data rather than run searches. For everyday case searching, use CourtListener or another case-law database.

### Is Google Scholar reliable for case law research?

It is reliable for finding cases, not for final validation. Google says its case law covers US state appellate and supreme court opinions since 1950, federal courts since 1923 and the Supreme Court since 1791. But its help page says Google "does not warrant that the information is complete or accurate," and it has no treatment citator showing whether a case was reversed or overruled. Use Scholar to locate and read opinions, then confirm each authority in a citator before you cite it.

### What is a citator?

A citator is a tool that tells you whether an authority is still good law and lists the later sources that cite it. Westlaw's is KeyCite, which Thomson Reuters says shows citing references, history and alerts. Lexis's is Shepard's, with signals for seven kinds of treatment. Bar-benefit Fastcase uses the Cert Citator, which vLex says reveals when cases have been overturned or criticized. On Westlaw, a red KeyCite flag means a case is no longer good law for at least one point of law. Free sites generally lack this feature.

### Which free legal sources are official?

Few of them. Govinfo's signed PDFs carry the "Authenticated U.S. Government Information" mark. Under [44 U.S.C. § 1507](https://www.law.cornell.edu/uscode/text/44/1507), the contents of the Federal Register "shall be judicially noticed." Under [1 U.S.C. § 204](https://www.law.cornell.edu/uscode/text/1/204), U.S. Code titles enacted as positive law are "legal evidence," while the others are prima facie evidence. The eCFR is an editorial compilation and is not yet official. Aggregators such as Google Scholar, Justia and Cornell LII are convenience copies, so cite the official source.

### How much does PACER cost?

PACER charges [$0.10 per page](https://pacer.uscourts.gov/pacer-pricing-how-fees-work), capped at $3 per document for items such as dockets, motions and briefs. The cap does not cover everything: searches, non-case-specific reports and transcripts have no maximum fee. Audio files cost $2.40 each. If you spend $30 or less in a quarter, fees are waived. Court opinions are always free, and parties get one free copy of documents through their electronic filing notice. RECAP shares documents other users already bought.

### How do I follow an MDL docket?

Start with the Judicial Panel on Multidistrict Litigation. Its [pending MDL report dated September 30, 2026](https://www.jpml.uscourts.gov/sites/jpml/files/Pending_MDL_Dockets_By_Circuit-September-30-2026.pdf) lists 162 MDL dockets, and it also publishes a [quarterly caseload summary](https://www.jpml.uscourts.gov/mdl-quarterly-caseload-summary). Then go to the transferee court's MDL page for orders and schedules. Set PACER or RECAP docket alerts so new filings reach you. Confirm status on the court's own record rather than relying on news coverage or a third-party tracker.

### Can AI legal research tools be trusted?

Not without verification. A peer-reviewed study by [Magesh and colleagues](https://arxiv.org/abs/2405.20362) found that leading AI research tools hallucinated 17% to 33% of the time, but those were 2024 versions, not the products sold today. ABA Formal Opinion 512 says relying on AI output without an appropriate degree of independent verification could violate the duty of competence under Model Rule 1.1. Grounded tools that link to sources are safer than general chatbots, but every answer still needs the six-step check before it reaches a filing.

### Do lawyers have to tell clients they use AI?

Not always, but sometimes. ABA Formal Opinion 512 says a client's informed consent is required before a lawyer inputs information relating to the representation into a self-learning AI tool, and boilerplate engagement-letter language is not enough. Where confidentiality rules do not require consent, the lawyer must still consider whether Model Rule 1.4 calls for disclosure, which is a case-by-case question. States adopt and interpret these rules differently. This is general information, not legal advice.

### Can a firm bill clients for legal research costs?

Usually not as a separate charge, unless the client agreed in advance. Under ABA Formal Opinion 93-379, ordinary library and research costs are overhead built into the lawyer's fee absent advance disclosure. ABA 512 adds that lawyers may not bill general overhead such as "maintaining a library," hourly billers must bill actual time, and a lawyer may not charge a client to learn an AI tool used regularly. Check the fee agreement and your state's rules first. This is general information, not legal advice.

### What research tools do personal injury lawyers need?

A personal injury practice needs a core platform or bar benefit with a citator, plus state-law and deadline resources for every state where it files. It also needs verdict and settlement data to value cases; LexisNexis, for example, lists a [Verdict & Settlement Analyzer and the CourtLink docket database](https://www.lexisnexis.com/en-us/solutions/legal-research.page). Docket access through PACER or RECAP covers federal matters. Product cases add regulator data: FDA's AEMS dashboard (formerly FAERS) for drugs and MAUDE for devices, read with their stated limits.

### What is the difference between legal research and litigation analytics?

Legal research finds and interprets the law; litigation analytics mines case data for patterns. LexisNexis puts it this way: traditional research "focuses on finding and interpreting statutes, case law, and legal commentary," while litigation analytics "analyzes historical case data and dockets to identify trends, judge behavior, settlement patterns, and outcome probabilities." Research tells you what the law is. Analytics helps you predict how a judge, court or opponent is likely to act. Most firms buy analytics as a separate product.