Compulife v. Newman — Trade Secrets

Q: Can scraping publicly available data be trade-secret misappropriation?

Yes, it can. In Compulife v. Newman the Eleventh Circuit held that using a bot to scrape an enormous volume of otherwise public quotes may be improper means of acquiring a trade secret, even though pulling individual quotes by hand would not be. The volume and method, not just the public availability, matter.

The Eleventh Circuit held that scraping a public database with a bot can be improper means of acquiring a trade secret, even when any single piece of the data is free to view.

A bot that pulls tens of millions of insurance quotes in four days does something no human ever could — and that gap, the Eleventh Circuit decided, can be the difference between fair competition and trade-secret theft. In Compulife Software, Inc. v. Newman, 959 F.3d 1288 (11th Cir. 2020), the court confronted a defendant who scraped a competitor’s quote engine that was, in a sense, open to the public, and held that the method and scale of the taking could make it “improper means” under trade-secret law. Authored by Circuit Judge Kevin C. Newsom, the decision has become a touchstone for the law of data scraping and, increasingly, for fights over how machines harvest information at scale.

At a glance

Case: Compulife Software, Inc. v. Newman, No. 18-12004 (11th Cir. May 20, 2020), 959 F.3d 1288.
Court: U.S. Court of Appeals for the Eleventh Circuit; opinion by Judge Kevin C. Newsom. On appeal from the U.S. District Court for the Southern District of Florida, where a magistrate judge tried the case by consent.
Posture: Appeal from a post-bench-trial judgment rejecting copyright and trade-secret claims; the Eleventh Circuit vacated and remanded.
Holding: Acquiring otherwise public data can constitute misappropriation by improper means; using a bot to scrape an infeasibly large volume of quotes from a competitor’s database may be improper means even though manually pulling individual quotes would not be.
Significance: A leading authority that automated, mass scraping of public-facing data can be trade-secret misappropriation, distinct from hacking or contract breach.

The quote engine and the scrape

Compulife develops and licenses a computerized system that calculates, organizes, and compares life-insurance quotes, selling insurers and agents access to a database that lets them generate up-to-date premium comparisons across carriers. The raw inputs — carriers’ published rates — are public, but Compulife’s value lies in the proprietary methodology and the organized database that turns those inputs into instant, comparative quotes. Compulife guarded that database as a trade secret while also exposing a quote function to the web.

The defendants, competitors operating insurance-marketing websites, wanted the same capability without paying for it. They deployed a bot that systematically queried Compulife’s public-facing quote engine across every combination of demographic variables, harvesting more than 40 million quotes in roughly four days — a feat the court observed would have taken a human thousands of hours. They then used the scraped data to populate their own sites. Compulife sued for, among other things, copyright infringement and trade-secret misappropriation under the Florida Uniform Trade Secrets Act. After a bench trial, the magistrate judge ruled against Compulife on both, reasoning in part that because the quotes were publicly accessible, scraping them could not be misappropriation.

Public access does not equal fair game

The Eleventh Circuit rejected that reasoning as a matter of law. Trade-secret law has long recognized that the manner of acquisition can be improper even when the information is, in principle, discoverable. The court invoked the canonical example from E.I. duPont de Nemours & Co. v. Christopher, where aerial photography of an unfinished plant — taken from public airspace — was held to be improper means of acquiring trade secrets. The lesson is that “improper means” turns on the conduct, not solely on whether the information could theoretically be obtained legitimately.

Applying that principle, Judge Newsom drew the line that has made the case famous: “Although Compulife has plainly given the world implicit permission to access as many quotes as is humanly possible, a robot can collect more quotes than any human practicably could. So, while manually accessing quotes from Compulife’s database is unlikely ever to constitute improper means, using a bot to collect an otherwise infeasible amount of data may well be.” Public availability of any single quote did not immunize the wholesale, automated extraction of the entire dataset. The court vacated the trade-secret ruling and remanded for the trial court to reconsider whether the defendants’ bot-driven scraping crossed into improper means.

Compilations, secrets, and what the bot took

Underlying the improper-means holding is a point about what the trade secret was. The defendants argued there could be no secret because individual quotes were free to obtain. But the court treated the database — the organized, comprehensive compilation generated by Compulife’s proprietary formula — as the asset at stake, not any lone quote. A valuable compilation can qualify as a trade secret even when its component data points are individually public, because the secret resides in the aggregation, organization, and the methodology that produced it.

That framing is what gives the scraping holding its teeth. By copying every possible quote, the bot effectively reconstructed Compulife’s compilation — appropriating the value Compulife had built — without ever breaching a password or signing a license. On remand, after a further bench trial, the district court found in Compulife’s favor on the trade-secret claim, awarded compensatory and punitive damages, and held the defendants jointly and severally liable; the Eleventh Circuit later affirmed that trade-secret ruling while sending the copyright questions back for additional findings. The through-line is that automated harvesting at superhuman scale can be actionable taking.

Open questions

How much volume is “infeasible”? The court tied impropriety to a scale no human could match but did not fix a threshold, leaving the line between permissible and improper scraping fact-specific.
Does a terms-of-service or technical barrier change the analysis? Compulife found improper means without relying on access restrictions; how robots.txt, CAPTCHAs, or contractual terms interact with the doctrine remains unsettled.
How far does the compilation theory extend? Treating an aggregation of public data as a secret invites disputes over when ordinary public information becomes a protectable compilation.

Implications

Scraping is not automatically lawful. Mass automated collection of public-facing data can be improper means of acquiring a trade secret, independent of hacking or contract breach.
Compilations can be secrets. The value in an organized database built from public inputs may itself be protectable, even when individual data points are not.
Scale and method matter. Courts will look at whether a bot did what no human practicably could, not merely at whether the data was viewable.
A theory for the AI era. Data owners increasingly invoke Compulife against large-scale scraping for training and aggregation, making it a frequent citation in modern data disputes.
Build defensible practices. Companies that scrape should weigh volume, method, and the target’s secrecy measures, because public visibility alone is not a safe harbor.

Frequently asked questions

Can scraping publicly available data be trade-secret misappropriation? Yes, it can. In Compulife v. Newman the Eleventh Circuit held that using a bot to scrape an enormous volume of otherwise public quotes may be improper means of acquiring a trade secret, even though pulling individual quotes by hand would not be. The volume and method, not just the public availability, matter.

What was the trade secret if the underlying numbers were public? The secret was Compulife’s database and the proprietary formula and methodology behind its life-insurance quotes, not any single quote. The court recognized that an organized, valuable compilation can be a trade secret even when its individual components are publicly accessible.

Why is this case important for AI and data scraping? It established that automated collection at a scale no human could achieve can constitute improper means under trade-secret law, giving data owners a theory against mass scraping that does not depend on hacking or breaching a contract. It is frequently cited in disputes over scraping and AI training data.