
Code of conduct and ethics training becomes useful when it moves beyond the annual signature and trains decisions: what do you say when a customer offers a gift, or when a boss pushes you to “make an exception”? A code that has been read and signed does not prepare you for these moments. A code that has been practiced does.
This article explains what the law expects of a code of conduct and why a signature does not change behavior. It also shows how to build role-plays around real dilemmas. It complements our guide to mandatory compliance training.
For companies subject to article 17 of the French Sapin II law, the code of conduct is one of the mandatory measures of the anti-corruption scheme. The recommendations of the French Anti-Corruption Agency (AFA) were published in the Journal officiel of January 12, 2021. They call for a code that defines and illustrates the behaviors to be prohibited.
The same recommendations ask for the code to be integrated into the internal rules, which triggers consultation of staff. They also specify that it must be understandable by non-specialists and organized by types of behavior. It draws on concrete cases from the risk mapping.
The code describes in particular clear prohibitions, such as gifts, conflicts of interest or facilitation payments. It also presents the whistleblower hotline, disciplinary sanctions and the person to contact in case of doubt. A code that stops at general principles, such as “act with integrity”, does not meet this expectation.
Outside Sapin II, many companies adopt a code of ethics by choice, for their culture or at their customers' request. The same principles of readability and examples apply.
The signature proves that the document was handed over. It does not prove that it is understood, still less that it will be applied under pressure. The AFA recommendations in fact acknowledge this: they ask for a record of the delivery of the code to be kept, and for knowledge to be checked during training.
The mechanism is simple. An ethical dilemma rarely arrives as a multiple-choice question. It comes in a conversation, with someone you know, who has a friendly argument and a deadline. The employee does not think “conflict of interest”, they think “I don’t want to offend this vendor”.
An online module trains the recognition of the right term. Only practicing the conversation trains the answer: the polite refusal, the question asked aloud, the request for approval from a third party.
The starting point remains the risk mapping. It says which roles are exposed to which temptations: buyers to gifts, salespeople to pressure over targets, project managers to personal ties. The most useful scenarios are those that fit these exposures, with the vocabulary and counterparts that the employee really meets.
Here are typical scenarios that come up in most companies. Each can be adapted by a trainer to their own roles.
Imagine a vendor who drops off a case of bottles in December, in the middle of a contract renegotiation. The employee must refuse or declare without causing offense. A possible sentence: “That’s very kind, but our code doesn’t allow me to accept it during a bid. I have to declare it.”
A service provider offers lunch, then an evening out. The employee learns to tell what remains proportionate from what becomes a quid pro quo. They practice asking their contact person before accepting: “I’ll get back to you after checking the internal rule.”
A manager asks to adjust a file to reach the quarter's target. The employee must say no to someone who has authority over them. “I can’t do it in this form. Can we look at another option together?” protects both the relationship and the rule.
An employee discovers that the candidate selected for a contract is a family member. They must declare it before going further, without waiting to be asked. The role-play rehearses the declaration to their manager, whose first reflex is often to play it down.
The AFA recommendations go in this direction: training of exposed staff relies on practical cases and scenarios linked to the risk mapping. It begins when the employee joins and continues regularly.
A good session is not judged by a single “right answer”. The trainer rather observes how the learner gets through the conversation. Four markers are useful.
These markers serve as a debrief grid. They make the session comparable from one participant to another, and from one session to the next for the same participant.
You must also accept imperfect answers. An employee who hesitates, justifies themselves too much or half gives in learns more than one who recites the rule. The second attempt is for correcting, not sanctioning.
A role-play gives the learner a counterpart who reacts to what they say. With Face Up, the trainer writes the scenario from their own cases: the persona can be a cordial vendor or a hurried boss. The learner answers aloud, then receives a debrief with quotes of their answers, their strengths and their points to improve.
The limit is clear: role-play trains the conversation, it does not replace the rule. In real cases, it is the code, the contact person and the hierarchy who decide. For an overview, see our page on role-plays.
A code of conduct touches several themes at once. Gifts and invitations fall under the Sapin II anti-corruption training. Respect for people joins harassment prevention. Data handling is found in the GDPR training for employees.
Rather than multiplying modules, a trainer can build a few cross-cutting scenarios. A single fifteen-minute conversation can combine a gift, a request for information about a third party and deadline pressure.
It is for companies subject to article 17 of the Sapin II law, which must also build it into the internal rules. Others can adopt one voluntarily. In every case, a short, illustrated code is better than an exhaustive text that nobody reads.
The AFA recommendations ask for a record of the delivery or availability of the code to be kept. A signature can serve as a record, but it does not replace training. Regular practice provides much stronger proof.
By agreeing with the employee on the difficulty of the situation, then having them try an answer. A dilemma where everyone is likeable is more instructive than a demonstration of what not to do.
Observe gift and conflict-of-interest declarations, the questions put to the contact person and the quality of answers in role-play. Our article on proof of training in an audit details the records to keep.
To build scenarios from your company's dilemmas, book a demo.
From reading to practice.
Pick a real-life situation and talk to an AI persona, by video with an avatar or by audio call. 5 to 10 minutes, no account, instant debrief.
Try a scenario →