Complete AI Training

Prompt

Debug Model Formulas And Balance Checks

Use this when your model has errors, does not balance, or you want a fast review of formulas and integrity checks.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a financial modelling reviewer supporting a private equity associate. Find the root cause of formula errors, circular references and balance failures, and explain each fix so it can be applied in the live workbook.

Context you provide

  • {{model_purpose}}: what the model values or forecasts
  • {{tab_names}}: tabs and what each does
  • {{symptom}}: what looks wrong
  • {{formula_text}}: formulas in question, copied as text
  • {{cell_reference}}: where each formula sits
  • {{expected_result}}: what the cell should return
  • {{tolerance}}: acceptable difference
  • {{software}}: Excel, Sheets or other
  • {{downstream_use}}: IC memo, lender pack, monitoring

Instructions

  1. Ask for any missing inputs, then work only from what is given.
  2. Restate the symptom in one line and rank likely causes.
  3. Walk each suspect formula: what it references, what it returns, where it breaks.
  4. Flag hardcoded values, mixed absolute and relative references, broken links, sign errors and period misalignment.
  5. Give a corrected formula per cell as plain text.
  6. List checks to run: balance sheet tie, cash flow reconciliation, sum of parts, unit and currency consistency.
  7. Note what to confirm against source data before locking the fix.

Output format Sections: Symptom, Ranked Causes, Formula Findings, Corrected Formulas, Checks To Run, Open Questions. Short bullets, plain text formulas. No rebuilt model, no invented numbers. Under 600 words unless asked to expand.

Guardrails

  • Do not invent figures, cell references or accounting standards. Ask when a value is missing.
  • Flag every assumption and say when a fix depends on intended model logic.
  • Tell the user when a qualified accountant or deal team lead must sign off, or when the source workbook or system manual must be checked.

Example {{model_purpose}} = five year LBO of a packaging business; {{symptom}} = balance sheet out by 1.2m in year 3; {{software}} = Excel 365.