The know-how you never wrote down

01 What the law actually protects

Ask a manufacturer who owns the knowledge of how their products are made, and the answer is immediate: the company, obviously. Ask a lawyer, and the answer starts with “it depends”, which in this case is not evasiveness. It depends on something specific, and it is worth knowing what.

Italian law protects company know-how mainly through article 98 of the Industrial Property Code, which implements the European directive on the protection of undisclosed know-how. It defines trade secrets as company information and technical-industrial experience which:

• a) are secret, meaning not generally known or readily accessible, as a body or in the precise configuration and combination of their elements, to experts and operators in the field;
• b) have economic value because they are secret;
• c) are subject, by the persons lawfully controlling them, to steps that can be considered reasonably adequate to keep them secret.

The three conditions are cumulative. Miss one and the information is not a trade secret, whatever it is worth and however obvious it seems that it belongs to the company. Article 99 then gives the holder the right to prohibit third parties from acquiring, disclosing or using it improperly, except where the third party arrived at it independently.

Condition (c) is the one that decides most cases, and it is the one nobody thinks about while the knowledge is being created.

02 You cannot defend a perimeter you never drew

“Reasonably adequate steps” sounds like a low bar, and in one sense it is: legal scholarship reading the case law is clear that the standard is a duty of diligence, not of result. The legislator does not require protection that cannot be broken by a determined expert. Passwords, access restrictions, confidentiality clauses, a policy on who may see what: this is the register.

But the requirement does something else before it does that. Taking steps to protect information forces you to say which information. And that identification, in the reading of the doctrine, is what draws the boundary of the right itself: protection covers only what the company identified in advance and actively demonstrated it intended to protect.

The Court of Brescia put the consequence plainly in a 2021 ruling, quoted in the legal literature: identifying the secret information makes it possible to establish its boundaries, which otherwise “might appear so blurred as to make it impossible to delimit the subject matter of the (alleged) right”. Not weak. Impossible.

03 Knowledge that exists only in a head

Here we have to be accurate, because the tempting simplification is wrong. The law does not require know-how to be written down. Article 98 protects “information” and “technical-industrial experience”, and the directive extends protection to objects and materials from which a secret can be inferred. Scholarship reading the provision concludes that the duty to identify does not imply a duty to embody the secret in documents: a trade secret can exist in unformalised company practice.

So a company whose finishing know-how lives in the head of one person is not automatically without rights. What it is without is proof. The same scholarship notes, with some understatement, that the absence of precise documentation instruments, know-how manuals and the like, may make proof of the right difficult and precarious.

Translate that into a Tuesday morning in a courtroom. The company has to state what the secret was, show it was not generally known in the sector, show it had economic value because it was secret, and show what concrete steps it took to keep it that way. For knowledge that was never written, never delimited, never restricted to anyone in particular, and that four other people in the workshop also partly held, each of those four things has to be reconstructed from testimony, years later, against a former employee who will say, accurately, that he simply knows his trade.

04 The fallback protection, and why it doesn’t catch this

There is a second line of defence. Even where information does not qualify as a trade secret, taking it can amount to unfair competition under article 2598 of the Civil Code. Italy’s Supreme Court has drawn the line, and it is worth reading in the Court’s own words.

In judgment 18772 of 12 July 2019, the Court of Cassation held that unfair competition can be established where a former employee of a competitor transfers a body of company information, even if it is not the subject of an industrial property right, but only where what is transferred is “an organised and structured body of cognitive data, even if not secret and protected, which exceeds the memory capacity and experience of the ordinary individual”.

Read that condition again from the perspective of a company that documented nothing. The fallback requires an organised and structured body of data. It protects you precisely when the knowledge had been assembled into something: a database, a structured archive, a body of material that goes beyond what a person can carry in their head. And its mirror image is the principle the same ruling establishes: knowledge that amounts to the employee’s own know-how and personal skills, acquired while working, does not found a claim at all. That, the law treats as theirs.

05 And the non-compete agreement?

The usual reply at this point is that the contract handles it. Sometimes it does, at a price and within limits that are stricter than most people remember.

During employment, article 2105 of the Civil Code imposes a duty of loyalty: the employee must not conduct business in competition with the employer, nor disclose information about the organisation and production methods of the business, nor use it in a way that could harm the business. But the settled reading is that the competition prohibition ends with the employment relationship. The confidentiality duty survives; the freedom to go and work for a competitor returns.

To restrict that, you need a non-compete agreement under article 2125, and the article is unforgiving. The agreement is void if it is not in writing, if no consideration is agreed in the employee’s favour, or if the restriction is not confined within defined limits of subject matter, time and place. Duration cannot exceed five years for executives and three years in other cases; agree more and it is cut back.

So the instrument exists, it costs money, it is time-limited, and it is routinely struck down for defects in drafting. It is also, by construction, the wrong tool for this problem: a non-compete can stop a person from working somewhere for three years. It cannot put back into the company anything they knew and never told anyone.

06 Why this matters more this year than last

None of the above is new law. What is new is how many people are about to walk out of the door with it.

In our study on the cost of retiring know-how we used the Excelsior forecast published by Unioncamere and the Ministry of Labour: in the wood and furniture supply chain there are 98 employees over 55 for every 100 under 35, the highest ratio of any industrial chain in Italy, and 30,000 people, 13.5% of the sector’s workforce, will have to be replaced by 2029. Most of those departures are retirements, which are not a legal problem: nobody takes finishing techniques to a competitor from a fishing boat.

But the same demographic squeeze makes experienced technical people the scarcest resource in the market, and scarce people get recruited. Every company in the district is trying to hire exactly the individuals who hold the knowledge of exactly its competitors. The legal question stops being theoretical the first time it happens to you, and by then the answer was decided years earlier, by whether anyone ever wrote anything down.

07 What to do about it, without a compliance project

The standard advice is a protection protocol: map the information you consider secret, apply access measures, review periodically. It is correct advice, and in a company of sixty people with no legal department it usually results in a document that is written once and never updated, because it asks people to describe knowledge in the abstract, which nobody has ever enjoyed doing.

The version that survives contact with a real workshop is narrower:

1. Write down what already gets asked. Not a manual. The questions colleagues actually ask each other, with the answers, accumulated in a shared place. This is the cheapest possible identification of the perimeter, because it is generated by ordinary work instead of by a project.
2. Keep it somewhere that has permissions. The measures required by the law are about who can see what. A shared drive where everybody sees everything is, on this specific point, worse than a smaller archive with explicit access rules.
3. Say who may see what, in writing, once. A short policy plus confidentiality clauses in employment contracts is the ordinary, unglamorous part, and it is the part that turns an archive into evidence.
4. Do it before the notice period. Knowledge is extracted from a person who is present and being asked good questions. Two weeks of handover at the end produce a tour of the building, not a transfer.

The point is not to win a hypothetical lawsuit. Litigation over trade secrets is slow, expensive and uncertain even when you are right. The point is that the same act that would give you something to defend also gives you something to use: an archive that answers questions is worth more on an ordinary Tuesday than in a courtroom you hope never to see. The legal protection is a by-product of documenting well. It just happens to be the by-product that makes the case to the owner.

08 Note on sources

This article is not legal advice. It describes how the rules are written and how courts have read them, for the benefit of people who run companies. Any concrete situation needs a lawyer who has seen the contracts.

• Legal texts are quoted from the current wording of article 98 and article 99 of the Industrial Property Code (Legislative Decree 30/2005) and articles 2105 and 2125 of the Civil Code. The European framework is Directive (EU) 2016/943, transposed in Italy by Legislative Decree 63/2018.
• The Cassation ruling is judgment 18772 of 12 July 2019, First Section, headnote Rv. 654770-03, quoted from the official headnote. The orientation was confirmed by judgment 3454 of 3 February 2022.
• The Court of Brescia quotation is taken from legal scholarship citing the 2021 ruling, not from the full text of the order, and is presented as such.
• The reading of “reasonably adequate steps”, including the point that documents are not legally required, comes from academic commentary on article 98(1)(c) and is attributed to scholarship rather than to the statute, because that is what it is.
• The demographic figures come from Excelsior, medium-term forecast 2025-2029, tables 3.3 and 3.4, already verified in our August 2026 study.

One thing we deliberately did not claim: that undocumented know-how has no protection at all. That would be false, and the sources say the opposite. What they say, and what this article reports, is that proof becomes difficult and precarious, and that the fallback route through unfair competition requires the organised body of data the company never built.

Sources

1. Codice della Proprietà Industriale, Legislative Decree 30/2005, articles 98 and 99 (current wording).
2. Codice Civile, articles 2105 and 2125.
3. Directive (EU) 2016/943 on the protection of undisclosed know-how and business information, transposed by Legislative Decree 63/2018.
4. Corte di Cassazione, First Civil Section, judgment 18772 of 12 July 2019 (Rv. 654770-03); Sixth Section, judgment 3454 of 3 February 2022.
5. Tribunale di Brescia, order of 23 April 2021, as cited in legal scholarship.
6. Academic commentary on the “reasonably adequate steps” requirement under art. 98(1)(c) CPI, Rivista di Diritto Societario.
7. Unioncamere, Ministry of Labour, Excelsior Information System, Previsioni dei fabbisogni occupazionali 2025-2029, 29 August 2025.
8. ekory, What the retiring craft is worth, August 2026.