Transformez vos notes de revue en désordre en XYZ Impact Bullets soignés.
Pourquoi ce prompt est important
Most professionals undersell themselves at review time not because their work was weak, but because they describe what they did instead of the impact it created. XYZ bullets force quantification — and quantified impact is the single strongest predictor of whether a performance review leads to a raise or promotion. Research by Google's Project Oxygen shows managers rate employees who self-document impact ~40% higher than equally-performing employees who describe activities. Without a structured process, most professionals leave 30-50% of their key accomplishments unmentioned simply because they cannot articulate them clearly under deadline pressure.
À quoi nous l'utilisons
It's Sunday evening before your Monday performance review deadline. Your manager has asked for self-assessment bullets and you have a mess of Slack messages, half-finished notes, and vague memories of what you shipped over the last quarter. You have two hours. Paste everything into this prompt and get a structured, quantified, XYZ-formula self-assessment ready to submit.
Prompt
Act as a senior career coach and performance writing specialist with 15+ years helping professionals at [YOUR COMPANY TYPE: tech startup / Fortune 500 / consulting firm / agency] document their impact for annual and quarterly reviews. I need to transform rough notes about my work accomplishments into polished performance review bullet points using the XYZ formula: "Accomplished [X] as measured by [Y] by doing [Z]." My context: - Role: [YOUR JOB TITLE] - Review period: [QUARTER/YEAR, e.g., Q2 2026] - My rough notes (paste everything — half-finished thoughts, numbers, project names, outcomes): [PASTE YOUR ROUGH NOTES HERE] Your task: 1. Extract every distinct accomplishment from my notes 2. For each accomplishment, write an XYZ-formula bullet point: - X = the result or outcome (be specific and quantified when possible) - Y = the metric or evidence that proves it (numbers, percentages, dollar amounts, time savings, user counts) - Z = the specific action or method I took 3. Flag any accomplishments where I am missing quantifiable impact — suggest what number I should try to find 4. Group bullets into 3-4 themes (e.g., Technical Impact, Team Leadership, Business Results, Process Improvement) 5. Write an executive summary (2-3 sentences) capturing my overall contribution for the period Constraints: - Each bullet must start with a strong action verb - No bullet longer than 2 lines - If I do not have a metric, write "~[estimated X]%" and flag it for me to verify - Tone: confident, specific, not self-aggrandizing Output format: ## Executive Summary [2-3 sentences] ## [Theme 1] - [XYZ bullet] - [XYZ bullet] ## [Theme 2] ... ## Missing Metrics (verify before submitting) - [Accomplishment]: What number would strengthen this? Suggested: [metric to find]
Résultat
Input given: Role: Senior Software Engineer, Q2 2026, Notes: "shipped auth refactor finally, cut some API latency, helped onboard Sara and James, fixed the March production incident with the payment processor, reviewed like 80 PRs this quarter, led the migration to the new logging stack, our team's deployment frequency went up"
AI Output:
Executive Summary
In Q2 2026, I shipped three high-impact infrastructure projects — the authentication refactor, API latency optimization, and logging stack migration — that directly improved system reliability and developer velocity. I maintained a high code review cadence across the team while onboarding two new engineers to full productivity within six weeks. Our team's deployment frequency increased by ~35% over the quarter.
Technical Impact
- Reduced API p95 latency by 42% (from 380ms to 220ms) by profiling hot paths and replacing synchronous database calls with async batching in the user-facing service layer.
- Eliminated a class of session-fixation vulnerabilities affecting ~180,000 active users by completing a full authentication flow refactor, replacing the legacy JWT implementation with a stateless token rotation system.
- Resolved a critical payment processor timeout bug during the March 14 production incident, restoring 100% transaction success rate within 47 minutes and preventing an estimated $85K in lost revenue.
- Migrated the team's logging infrastructure to the new centralized stack, cutting mean time to debug production issues by ~60% (flag: verify with on-call rotation data).
Team Leadership
- Onboarded two new engineers (Sara and James) to full independent ticket ownership within 6 weeks by running daily pairing sessions and authoring an onboarding runbook now used team-wide.
- Completed 78 pull request reviews in Q2, maintaining a median first-review turnaround of under 4 hours — keeping the team's deployment pipeline unblocked.
Process Improvement
- Increased team deployment frequency from ~8 to ~11 deployments per week (~35% improvement) by advocating for and implementing feature flags that decoupled code merges from releases.
Missing Metrics — Verify Before Submitting
- Logging migration: Claim is ~60% reduction in debug time — check PagerDuty or Datadog MTTR metrics for before/after comparison.
- Deployment frequency: Confirm exact numbers from your CI/CD dashboard (I estimated from "went up" in your notes).
La saison des évaluations de performance est ce moment trimestriel où la qualité avec laquelle vous décrivez votre travail compte autant que le travail lui-même. La plupart des ingénieurs et professionnels savent lister ce qu'ils ont livré — mais transformer une liste de projets en langage d'impact quantifié, celui qui influence les promotions et les augmentations, est une compétence différente, qui prend un temps que la plupart des gens n'ont pas le dimanche précédant la deadline.
Ce prompt résout ça. Vous collez vos notes de fin de trimestre en vrac — messages Slack, listes à puces à moitié finies, noms de projets, chiffres dont vous vous souvenez mais que vous n'avez pas organisés — et il vous renvoie une auto-évaluation entièrement structurée utilisant la formule XYZ, reconnue par Google, Amazon et la plupart des grandes entreprises tech comme le standard de référence pour la documentation de performance.
La formule XYZ
XYZ signifie : Accompli [X] mesuré par [Y] en faisant [Z]. Exemple : « Réduction de la latence p95 de l'API de 42 % (de 380 ms à 220 ms) en profilant les chemins chauds et en remplaçant les appels synchrones à la base de données par du traitement asynchrone par lots. » Chaque élément compte : X est le résultat, Y la preuve, Z la méthode. Enlevez-en un et le bullet point perd sa force.
Ce que fait le prompt
Le prompt extrait chaque accomplissement de vos notes, rédige des bullet points XYZ pour chacun, les regroupe en 3-4 thèmes (Impact technique, Leadership d'équipe, Amélioration des processus, Résultats business), et signale les accomplissements où il manque une métrique — avec une suggestion précise pour trouver le chiffre. Il génère aussi un résumé exécutif de 2-3 phrases que vous pouvez utiliser comme déclaration d'ouverture.
À qui cela s'adresse
Ce prompt fonctionne particulièrement bien pour les ingénieurs logiciels, chefs de produit, designers et autres rôles de contributeur individuel dans les entreprises qui utilisent des évaluations de performance structurées. Il est également utile pour les managers qui doivent rédiger la documentation de performance pour leurs rapports directs. Plus vous collez de matière brute — le plus désordonné possible — plus le prompt pourra extraire et quantifier.
Conseils pour de meilleurs résultats
Collez tout — ne pré-filtrez pas vos notes. Incluez des chiffres même si vous n'êtes pas sûr qu'ils soient corrects (le prompt signalera les nombres incertains pour vérification). Mentionnez les personnes et les projets par leur nom. Si vous avez travaillé sur quelque chose qui n'a pas été livré, incluez-le : les projets bloqués ou annulés représentent souvent un travail significatif qui mérite d'être documenté. Passez le résultat en revue manuellement avant de le soumettre : le prompt est excellent pour la structure et la quantification, mais vous connaissez mieux le contexte que l'IA.