Ten Questions to Ask a Legal Data Vendor Before You Buy

Short answer: before you pick a legal data vendor, ask ten things. Where does each document come from? Is it the official text? How current is it? Can you read it as of a past date? How are gaps reported? What does the licence allow? How are changes announced? How are citations matched? How is customer data handled? What happens to you if the vendor changes or stops? A good vendor answers each one in plain words and shows you evidence.

TL;DR

• A legal data vendor (also called a legal data provider) sells laws and regulations as data that your product or team can use.
• You are buying trust as much as text. If one section is wrong, stale or missing, your customer sees your name on the mistake.
• Ask for evidence. A link to the official source for any section beats a promise.
• The questions below each come with what a good answer sounds like and what a weak one sounds like.
• A one-page table at the end summarises all ten.

Why this list exists

If you are building a product on top of law, such as a compliance tool or a contract tool, the law is your raw material. A supplier who is slow, vague or wrong becomes your problem. Use these questions in a first call, in a written questionnaire, or in a trial.

A short history of getting the law wrong

The government itself learned this the hard way. In December 1934 the US Supreme Court heard Panama Refining Co. v. Ryan, the first case to reach the Court that challenged the National Industrial Recovery Act, the centerpiece of the New Deal (President Roosevelt's economic program). It was about a rule for the oil industry. Then it came out that the paragraph being attacked had been taken out of the oil code by an executive order on September 13, 1933, before the lawsuits were even filed. Another order put it back on September 25, 1934. Meanwhile nearly everyone, lawyers and judges included, was working from the old text.

When the Court decided the case on January 7, 1935, it said the persons affected, "the prosecuting authorities, and the courts, were alike ignorant of the alteration," and that the attack "was upon a provision which did not exist."

That embarrassment is a big reason the Federal Register exists. Congress passed the Federal Register Act on July 26, 1935, and the first sixteen-page issue appeared on March 14, 1936, according to the National Archives' own history. More than ninety years later a buyer's questions are the same two: is this the version in force, and how would I know? Questions 3, 4 and 7 below are that story turned into a checklist.

How to evaluate a legal data vendor: the ten questions

1. Where does each document come from?

Why it matters: a copy of a copy collects errors, and you cannot fix an error if you cannot find its origin.

A good answer names the official publisher for each body of law, such as a legislature, a state agency or the federal government, and shows a source link on each section.

A weak answer says "public sources" or "we aggregate from many places" and cannot name the publisher for a given state. It also shows why question 2 comes next.

2. Is the text the official version?

Why it matters: in some places only one version counts as law, and a convenient copy may differ from it.

A good answer says, for each source, whether the publisher calls it official. This varies. Idaho's code says the electronic administrative code is the official text, and South Dakota's code lists numbered volumes as its official code. The federal eCFR describes itself as an informational resource and the annual edition is the formal text. Our guides to Idaho and South Dakota show how different the rules are.

A weak answer treats "official" and "accurate" as the same thing.

3. How current is it, and how do I see that?

Why it matters: a law that changed last month is a different law.

A good answer gives a date for each section or source, such as "current through" or "last checked", and says how often each source is refreshed. Our note on data freshness explains what to look for.

A weak answer says "updated regularly" with no schedule and no date on the record.

4. Can I read a law as of a past date?

Why it matters: disputes, audits and contracts usually turn on what the law said at a past moment.

A good answer says yes or no plainly, then explains how the older versions are made. Are they stored copies of what was published at the time, or are they rebuilt from amendment notes? The first is firmer evidence. Ask for a worked example. Texas data breach law, Tex. Bus. & Com. Code § 521.053, was added in 2007 and has been amended six times since, most recently in 2023, so ask the vendor to show you the 2020 text next to today's. The legislature's own history note lists the amendments as 2009, 2011, 2013, 2019, 2021 and 2023, and the original took effect on April 1, 2009. For a date in mid-2020, two changes sit between that text and today's, one effective September 1, 2021 and one effective September 1, 2023. For federal rules, see how annual CFR editions work.

A weak answer says "we keep history" with no method.

5. How do you report gaps and missing jurisdictions?

Why it matters: an empty search result looks the same whether the law does not exist or the vendor does not hold it.

A good answer points to a coverage report you can read, by state and by type of law, and says what a "no result" means.

A weak answer claims "full coverage" and offers no list.

6. What does the licence let me do?

Why it matters: you may need to store the text, show it to your own customers, use it in an AI feature or resell it.

A good answer is written down and covers storing, displaying, training or fine-tuning a model, and passing text to your customers. It states what happens to the licence if you change business model.

A weak answer says "it's fine, everyone does that" or sends a standard form that is silent on display rights.

Free to read does not mean free to copy, and the line is not always where people expect. The US Supreme Court's 2020 ruling on Georgia's annotated code is a good example, and our buyer's guide to legal data tells the story.

7. How are changes detected and announced?

Why it matters: if you serve compliance answers, a missed amendment is a wrong answer.

A good answer describes how the vendor finds a change, how long that takes, and how you hear about it, for example by an alert you can subscribe to for a given section.

A weak answer puts the burden on you to re-download everything and compare.

8. How do you match a citation to a section?

Why it matters: the same string can name different sections in different states, and users type citations in many shapes. If you are new to citations, read how to cite a statute first.

A good answer explains what happens with an ambiguous cite, a cite to a repealed section and a cite that matches nothing. The right result for a cite that matches nothing is a clear "not found".

A weak answer always returns something, even a guess.

A guess is not harmless. In June 2023 a federal judge in New York penalized two lawyers and their firm $5,000 in Mata v. Avianca after they filed court papers that cited opinions ChatGPT had invented, complete with fake quotes. The judge was even-handed about the tools: "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." A lookup that hands back a plausible section when it should say "not found" feeds the same problem.

9. How do you handle security reviews and customer data?

Why it matters: your own buyers will ask you about your suppliers.

A good answer has a security document ready, says what customer data they store (queries, account details, logs), for how long, and who can see it, and tells you how to run a review.

A weak answer says "we take security seriously" and nothing else.

10. What happens to me if you change or stop?

Why it matters: your product depends on this feed.

A good answer covers notice before a change, how long old endpoints keep working, how you can export what you have rights to, and what the contract says on termination.

A weak answer has no notice period and no exit terms.

What "official" and "current" mean for the US Code

Even the federal government's own book has two levels, which is a handy way to see why questions 2 and 3 are worth asking. The Office of the Law Revision Counsel (the House of Representatives office that prepares the United States Code) explains on its site that under federal law the printed Code "establishes the law prima facie." That is Latin for "at first look": it counts as correct unless someone shows otherwise. The exception is the titles Congress has enacted into positive law, meaning Congress voted on the Code text itself. For those titles the text is "legal evidence of the law."

Currency has its own wrinkle. The same page says "there is necessarily a lag between the date a law is enacted and the date on which it appears in the printed version of the Code." The online version is updated on an ongoing basis, and the office keeps a separate page that says how current it is. That plain "current through here" statement is what a good vendor shows you on each record.

Questions to ask yourself too

Before the call, write down which states and types of law you need, what past dates matter, and who inside your company will review the licence. A vendor can only answer what you ask. Our guide to types of primary law is a quick way to list what you need.

How to use this list

Send the ten questions in writing and ask for answers in writing. Then score each answer as good, partial or weak, and ask for evidence on every "good". Test two or three answers yourself: open a section, follow its source link and compare the text with the official site. Vaquill AI is one supplier you can run this list against, and so is every other. A vendor who answers all ten with specifics is rarer than one with a polished demo.

The ten at a glance

This guide is part of US Law Data: The Complete Guide, a map of where US law comes from and how to use it.

FAQ

What is a legal data vendor? A legal data vendor is a company that supplies laws, regulations and sometimes court decisions as data, usually through an interface that software can call. Buyers include legal technology companies, compliance teams and AI builders.

Why not just use free government websites? Official sites are the best source for any single document. They are made for reading one page at a time, and their formats, update times and structure differ from state to state. A vendor does that gathering and clean-up for you. The ten questions tell you whether it was done well.

How long should a vendor due diligence take? For a small product team, a written questionnaire and one trial of a few days is usually enough. Add time if your own customers will audit you.

Is a bigger vendor safer? Size does not answer these questions. A small vendor with clear sources and written terms can beat a large one with vague answers. Judge by evidence.

Which question matters most? It depends on what you build. A compliance product leans on questions 3, 4 and 7. A product that shows law to end users leans on 2 and 6. If you can only ask one, ask for a worked example of question 4.

Should I ask for a trial? Yes. Pick five sections you know well, from different states, and check each against the official site. Our piece on where legal data comes from explains what to compare.

For the data behind this checklist, see the US primary law API.

Priyansh Khodiyar

Co-Founder & CTO

Priyansh leads engineering and AI at Vaquill AI: the pipelines that pull statutes, regulations and court rules from every US jurisdiction's official publisher, and the REST API, MCP server and open dataset that serve them.