Annual Performance Review Roleplay Simulator: Prepare Managers and Employees

The annual review is often treated as an obligation: a form to fill in, an hour squeezed between two urgent tasks. Yet it is when an employee hears what their manager thinks of their year and their future. Handled badly, it leaves marks.

• A vague assessment: the employee leaves not knowing where they stand.
• A surprise: a criticism never voiced lands all at once, and trust erodes.
• An empty promise: training or a raise mentioned with no framework, then forgotten.
• Unverifiable goals, which bring back the same discussion twelve months later.

What managers lack is rehearsal: announcing a disappointing rating out loud, holding a position when asked for a raise. A virtual counterpart provides that testing ground without exposing anyone.

In AI-Coaching, the counterpart is a persona you configure (role, personality, AI profile). The session is played in text or voice; an AI observer coaches you discreetly in real time, then produces a detailed debrief. You can simulate a review from start to finish or a single stage.

Opening and tone

The manager restates the purpose, the structure and the time available, then lets the employee speak first about their year. Avoid the administrative opener or the offhand remark.

Review of the year

It rests on dated facts and achievements, not impressions. You balance what worked with what must improve, and practice listening: rephrase the employee’s reading of the year before giving your own.

Goals and development plan

Goals for the coming year, training needs, expected support: each goal must be specific, measurable and dated, and each commitment from the manager realistic.

Closing and next steps

Oral summary, each side’s commitments, a date for a checkpoint. A review that ends with “let’s talk about it later” is one to replay.

Each scenario becomes a mission with a configured counterpart; adapt context, difficulty and vocabulary to your company.

1. The positive review

Counterpart: pleased, but eager to know “what next?”. Success: cite specific achievements, then open the conversation on growth. Common mistake: rushing through because “everything is fine”.

2. The disappointing rating

Counterpart: surprised, then defensive or hurt. Success: state the rating clearly, back it with examples, welcome the reaction without backing down. Common mistake: burying the message under compliments.

3. The refused raise request

Counterpart: well argued, comparing with the market. Success: say no clearly, explain what was decided and by whom, explore what remains possible without promising anything you do not control. Common mistake: “it is not up to me”.

4. SMART goals

Counterpart: cooperative but vague, or worried about unrealistic targets. Success: agree on specific, measurable, achievable, relevant and time-bound goals, built with the employee. The demo includes a scenario called “Échange collaborateur SMART : projet portail client” (a SMART goal-setting conversation) to try this. Common mistake: imposing numbers without discussing the means.

5. The demotivated employee

Counterpart: short answers, eyes elsewhere. Success: describe what you observe without interpreting, ask open questions, accept silences. Common mistake: offering a solution before understanding the cause.

For tenser exchanges (reprimands, conflict), see the difficult conversations simulator for managers.

An employee who arrives without having prepared their assessment or requests leaves the manager to run the whole conversation. With the same tool, they practice:

• presenting their year in three achievements;
• asking for a raise or a new role with facts;
• receiving criticism without bristling, by asking for examples;
• voicing disagreement calmly, with a proposed next step.

The persona is configurable: you create a demanding, rushed or supportive manager, and the employee plays themselves. The demo includes, for example, situations facing a manager (“Éthique : contournement de règles internes”, about bypassing internal rules) or an HR exchange (“RH : employé manquant de ponctualité”, about punctuality).

• The annual performance review is a company practice: review of the year, goals, performance. Your organization sets its format.
• The professional interview (entretien professionnel) comes from the French Labor Code. It covers career prospects, skills and training; under article L6315-1, it does not cover the evaluation of the employee’s work.

The framework has just changed. On Légifrance, article L6315-1 in force on December 31, 2025 (law no. 2025-1403 of December 30, 2025) refers to a “career path interview”: one in the first year after hiring, then every 4 years, with a comprehensive review every 8 years. Transitional provisions apply to collective agreements; have HR or a legal adviser check your situation.

In simulation, a manager who blends the two blurs the message: you practice moving clearly from the “evaluation” phase to the “career and skills” phase.

The AI observer follows the exchange, then produces a debrief against configurable criteria. Here is a base to adapt with your HR team; score each criterion from 1 to 4 using observable signals.

Your review form, management charter or skills framework can be fed into the mission (RAG) so the debrief sounds like your organization.

1. Eight weeks before. Pick the scenarios, validate the grid with HR, enrich the missions with your documents.
2. Six weeks before. Pilot with volunteer managers, adjust the difficulty.
3. Four weeks before. Open a pathway to managers, through a secure link (expiry, usage limit) or the platform.
4. Two weeks before. Two or three short sessions per manager, each followed by a chat with a peer or coach about one passage of the debrief.
5. During the cycle. The same pathway for employees, for preparation.
6. After. Collect feedback and adjust the scenarios.

You group scenarios into pathways and playlists, track progress, and distinct profiles (operator, editor, user) keep design and practice apart. These conversations touch on internal topics: cloud or local installation, and a choice of language model (Mistral, OpenAI, Azure, Gemini, deepseek, self-hosted) let you adjust control over your data. Define who sees what before launch, with HR, your data protection officer and employee representatives.

For the overall approach, read training managers, preparing managers and coaching with AI and the page on the AI simulator for internal training. To scope a project with your own form, request a demo.