Protecting Confidential Information — Trade Secrets & Proprietary Methodology

A Data Scientist Spent Four Years Building a Proprietary Sales Algorithm at Their Previous Employer. Their New Employer Hired Them to Build the Same Thing. No Files Were Taken — Everything Is in Their Head. Is That a Trade Secret Violation?

A real trade secret and post-employment compliance scenario — with three decision options and the right answer.

The Situation

A quantitative analyst spent four years at a financial services firm building a proprietary sales prioritization model — an algorithm that scores inbound leads and routes them to the appropriate sales tier based on 40 variables. The model was developed using the firm’s proprietary client data, transaction history, and behavioral patterns. The firm’s close rate on model-routed leads is three times the industry average. The algorithm is a significant competitive advantage and the firm treats it as highly confidential.

A direct competitor recruits the analyst at a 40% salary increase, specifically to build a similar lead scoring capability. The analyst takes no files. They download nothing. They leave on good terms. Six months after joining the new company they have built a functionally identical model from memory — the same variable selection approach, the same weighting methodology, the same calibration logic. The original firm’s sales performance advantage disappears within two quarters. Their data science team notices and outside counsel gets involved.

Choice ABuild the model — it is the analyst’s expertise, not their former employer’s property. The analyst developed the methodology through their own intelligence and effort over four years. Their skills and knowledge are not the firm’s intellectual property. No files were taken and the new model will be built using entirely new data.

Choice BBefore beginning any model development at the new company, disclose to the new employer’s Legal team that the role requires building a capability that is substantially similar to a proprietary methodology developed at a previous employer — and ask for guidance on how to proceed in a way that does not misappropriate the former employer’s trade secrets.

Choice CBuild a model that is similar in function but different in architecture — using different variable categories and a different weighting approach — so it cannot be characterized as a copy of the original. Draw the line independently based on what feels sufficiently different.

The Right Call

Choice B — Involve Legal before beginning any model development.

Choice C is the most common response from technically sophisticated employees and one of the most legally dangerous. An analyst who draws the line themselves — deciding what is sufficiently different from the original — is making a legal judgment they are not qualified to make, in a situation where they have an obvious incentive to conclude the new model is acceptable. Courts do not assess trade secret misappropriation solely based on architectural similarity. They assess whether the person who built the new system had access to the original trade secret and whether that access influenced the design — regardless of surface differences. Choice B involves the people qualified to make that assessment before the exposure is created.

“I built it — it’s my expertise” is the most convincing wrong answer in trade secret law.

There is a genuine and important distinction between general skills and specific proprietary methodology. An analyst’s ability to build predictive models, select variables, and calibrate weights is a general expertise they bring to every employer. The specific combination of variables, the proprietary weighting logic developed over four years of iteration on confidential client data, and the calibration approach optimized for that firm’s specific market segment are not general expertise — they are trade secrets developed using the employer’s resources, data, and time. That distinction is not intuitive. It requires Legal’s assessment, not the analyst’s.

“No files were taken” does not mean no trade secret was misappropriated.

This is the central misconception in cases involving memorized methodology. Courts have consistently held that trade secret misappropriation through memory is actionable under the DTSA and state equivalents. The absence of a download log or file transfer record makes the case harder to prove — it does not make the misappropriation legal. An analyst who carried a proprietary algorithm in their head and reconstructed it at a competitor has taken something that didn’t belong to them, regardless of the storage medium.

The new employer, which was hired specifically to replicate the capability, has its own exposure.

A company that recruits an analyst from a known competitor specifically to build a capability that competitor is known to have has created an inference of intentional misappropriation that courts examine carefully. If the job description, the recruiting conversation, or early internal communications suggest the new employer wanted the proprietary methodology and not just the analyst’s general skills, the receiving company’s exposure under the DTSA is significant. This is why a well-run Legal team at the receiving company would want this conversation before the role starts, not after the model is deployed.

The analyst’s good faith is relevant to damages — not to liability.

An analyst who genuinely believed their methodology was their own expertise, rather than their former employer’s trade secret, acted in good faith. That good faith affects the assessment of punitive damages and, potentially, the injunctive relief analysis. It does not eliminate the misappropriation claim or the receiving company’s liability. Trade secret law does not require bad intent for liability — it requires the misappropriation of protected information, however it occurred.

What is the legal distinction between general professional expertise and a trade secret methodology?

General expertise includes skills, knowledge, and techniques that a professional develops over their career and can apply at any employer — statistical modeling approaches, programming languages, industry knowledge, and analytical frameworks. Trade secret methodology is the specific proprietary application of those skills to a particular employer’s problem — the specific variables, the specific weighting logic, the calibration developed using that employer’s confidential data. The distinction is not always clear and that’s exactly why Legal’s assessment is required rather than the employee’s own judgment.

Is memorized trade secret information protected under the Defend Trade Secrets Act?

Yes. The DTSA explicitly covers information that is stored in memory as well as information in tangible form. Courts have applied trade secret protection to memorized customer lists, memorized formulas, memorized processes, and memorized algorithmic methodologies. The absence of a physical record makes cases harder to prove but does not eliminate the legal protection or the liability for misappropriation.

What should a new employer do before assigning a competitive hire to build a capability similar to one they built at their previous employer?

Involve Legal before the assignment begins. A structured assessment of what the employee built at their previous employer, what confidentiality obligations they are subject to, and whether the new role can be designed to use their general expertise without requiring them to recreate proprietary methodology is the responsible approach. Organizations like Hitachi that have formal competitive hire protocols — including a signed declaration that the new hire has not brought and will not use former employer confidential information — are protecting both themselves and the employee from a situation that can destroy careers and generate significant litigation costs.

What remedies are available to a company whose proprietary algorithm has been misappropriated by a former employee?

Under the DTSA, remedies include injunctive relief preventing the use of the misappropriated methodology, actual damages including lost profits and unjust enrichment, exemplary damages up to two times actual damages for willful misappropriation, and attorney’s fees. The injunction can require the competitor to stop using the methodology entirely — which in this scenario could mean shutting down a deployed sales system and the competitive advantage it created. The combination of injunctive relief and damages exposure makes algorithm misappropriation one of the highest-stakes areas of trade secret litigation.

How to Use This Scenario in Training

Designed for data scientists, engineers, quantitative analysts, researchers, and senior technical professionals — the audience most likely to carry proprietary methodology in their heads and least likely to have received compliance training that addresses their specific situation. Also highly relevant for hiring managers and HR business partners involved in recruiting from competitors.

The key recognition skills: understanding that “no files were taken” is not a legal defense; understanding that the distinction between general expertise and trade secret methodology requires Legal’s assessment; and understanding that a functionally similar model built by someone with access to the original is legally exposed regardless of architectural differences.

More Confidentiality & Trade Secret Scenarios

Want These Scenarios in Your Compliance Program?

Xcelus builds scenario-based trade secret and confidentiality training for technical professionals, data scientists, engineers, and the organizations that hire them competitively.

View the Compliance Reinforcement Kit →
Contact Xcelus