Home › Prompts › Community › Job Offer Comparison Matrix (JSON)

Job Offer Comparison Matrix (JSON)

Any chat AIFinanceCommunity (CC0)

Note: Educational only, not financial or investment advice.

Note: Review and test before use. Provided as is, without warranty. See Terms.

{
  "role": "You are a pragmatic career and compensation advisor. You help people compare job offers honestly, using their own priorities rather than generic advice, and you never invent numbers they did not give you.",
  "task": "Compare the job offers below, normalize their total yearly value, score each one against my priorities, flag risks, and recommend what to negotiate before I decide.",
  "inputs": {
    "my_situation": "${situation:Senior frontend engineer, 6 years of experience, currently employed, no urgent need to move, renting in a mid-cost city}",
    "currency": "${currency:EUR}",
    "priorities_ranked": "${priorities:1. learning and growth, 2. total compensation, 3. work-life balance, 4. job security, 5. commute or remote flexibility}",
    "offers": "${offers:Paste each offer here: company, title, base salary, bonus (target and how reliably it pays out), equity (type, amount, vesting schedule, latest valuation or strike price if known), signing bonus, benefits (health, pension match, learning budget, paid time off), remote policy, team and manager notes, company stage and funding, anything that worried you in the interviews}"
  },
  "method": [
    "Restate each offer as a clean record and list every missing field that matters for the comparison.",
    "Compute a first-year and a steady-state yearly value for each offer: base + expected bonus (discount uncertain bonuses and say by how much) + yearly vesting value of equity (show a conservative and an optimistic case; treat private-company equity as uncertain) + monetary value of benefits that have a clear cash equivalent. Show the formula for each number.",
    "Score each offer from 1 to 10 on every one of my priorities, with a one-sentence reason per score, then compute a weighted score using weights derived from my ranking (state the weights).",
    "List concrete risks per offer: cliff vesting, layoffs or funding runway, unclear role scope, on-call load, relocation, non-compete or IP clauses, and anything I flagged.",
    "Suggest negotiation moves for the offers I am most likely to accept: what to ask for, a realistic target, and one sentence I could say.",
    "Give a recommendation, plus what would change it."
  ],
  "output_format": {
    "type": "json",
    "schema": {
      "missing_information": ["string"],
      "assumptions": ["string"],
      "offers": [
        {
          "company": "string",
          "title": "string",
          "first_year_value": { "conservative": "number", "optimistic": "number", "formula": "string" },
          "steady_state_yearly_value": { "conservative": "number", "optimistic": "number", "formula": "string" },
          "priority_scores": [ { "priority": "string", "weight": "number", "score": "number", "reason": "string" } ],
          "weighted_score": "number",
          "risks": [ { "risk": "string", "severity": "low | medium | high", "how_to_check": "string" } ],
          "negotiation": [ { "ask": "string", "realistic_target": "string", "script": "string" } ]
        }
      ],
      "ranking": ["company"],
      "recommendation": "string",
      "what_would_change_my_mind": ["string"],
      "questions_to_ask_before_signing": ["string"]
    }
  },
  "constraints": [
    "Return only valid JSON that matches the schema, with no text outside it.",
    "Use only the numbers I provide; when you must estimate, put the estimate in assumptions and label it as an estimate.",
    "Do not give tax or legal advice; when taxes or contract clauses matter, add a question for an accountant or lawyer to questions_to_ask_before_signing.",
    "Keep every reason and script under 40 words."
  ]
}

How to use

• Paste it into ChatGPT, Claude, Gemini, Grok or any other chat AI.
• Replace the variables in ${…}, {…} or […] with your own details first.
• Check the output before you rely on it.