ChatGPT Prompts for Financial Analysis
Thirty structured prompts to analyze ratios, statements, trends, variances, forecasts, and valuations. ChatGPT is analysis support, not financial advice — always verify every number against your source data.
In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 15 prompts are free instantly — no signup needed. Hand-curated and tested by the AI Academy team.
Ratio Analysis
5 promptsLiquidity Ratio Breakdown
1/30<context> You are a financial analyst reviewing liquidity for [COMPANY]. Balance sheet figures: [FIGURES] (current assets, current liabilities, inventory, cash & equivalents, receivables). Reporting period: [PERIOD]. Industry: [INDUSTRY]. </context> <task> 1. Compute the current ratio, quick (acid-test) ratio, and cash ratio. Show the formula and the inputs used for each. 2. Present results in a table with columns: Ratio | Formula | Value | Interpretation. 3. Flag any ratio below 1.0 and explain the short-term solvency implication. 4. Note 2-3 limitations of these ratios (e.g. seasonality, off-balance-sheet items). 5. End with: "Verify all inputs and outputs against the audited financials before relying on these figures." </task>
A clear liquidity snapshot with formulas, computed values, and plain-language interpretation.
Pro tip: Paste the raw balance-sheet line items rather than pre-computed ratios so ChatGPT can check the arithmetic and you can audit its inputs.
Profitability Ratio Analysis
2/30<context> You are analyzing profitability for [COMPANY]. Income statement & balance sheet figures: [FIGURES] (revenue, COGS, operating income, net income, total assets, equity). Period: [PERIOD]. </context> <task> 1. Calculate gross margin, operating margin, net margin, ROA, and ROE. Show each formula and the numbers plugged in. 2. Output a table: Metric | Formula | Value. 3. Explain what each metric reveals about [COMPANY]'s earnings quality. 4. Identify which margin is strongest and weakest and offer a hypothesis for why. 5. List the data you would need to benchmark these against peers. 6. Remind me to verify every computed figure against the source statements. </task>
Margin and return metrics with formulas, an earnings-quality read, and benchmarking next steps.
Pro tip: Ask ChatGPT to recompute one ratio a second time as a self-check — LLMs occasionally slip on multi-step arithmetic.
Leverage & Solvency Ratios
3/30<context> Assess long-term solvency for [COMPANY]. Figures: [FIGURES] (total debt, total equity, EBIT, interest expense, total assets). Period: [PERIOD]. </context> <task> 1. Compute debt-to-equity, debt-to-assets, and interest coverage (EBIT / interest expense). 2. Show formulas and inputs, then present a table: Ratio | Value | What it signals. 3. Classify the leverage profile as conservative, moderate, or aggressive with reasoning. 4. Explain the risk if interest coverage falls below 1.5x. 5. State that these conclusions are analysis support only, not financial advice, and that figures must be verified. </task>
Leverage and coverage metrics with a risk-profile classification and verification reminder.
Pro tip: Tell ChatGPT your industry — acceptable leverage for a utility differs sharply from a SaaS company.
Efficiency / Activity Ratios
4/30<context> Evaluate operational efficiency for [COMPANY]. Figures: [FIGURES] (revenue, COGS, average inventory, average receivables, average payables, total assets). Period: [PERIOD]. </context> <task> 1. Calculate inventory turnover, receivables turnover (and DSO), payables turnover (and DPO), and asset turnover. 2. Show formulas and inputs for each; output a table. 3. Compute the cash conversion cycle (DSO + DIO - DPO) and explain what it implies for working capital. 4. Suggest 2-3 operational levers that could improve the weakest metric. 5. Note assumptions made about averages and ask me to confirm them. </task>
Turnover metrics plus a cash conversion cycle and working-capital interpretation.
Pro tip: Provide opening and closing balances so ChatGPT uses true period averages instead of point-in-time figures.
Ratio Trend Across Periods
5/30<context> You are comparing key ratios for [COMPANY] over multiple periods. Multi-period figures: [FIGURES] (provide the same line items for each period). Periods: [PERIOD]. </context> <task> 1. Recompute the core ratios (current, gross margin, net margin, ROE, debt-to-equity) for each period. 2. Present a matrix: Ratio | Period 1 | Period 2 | Period 3 | Trend (up/down/flat). 3. Highlight the two ratios with the most material movement and quantify the change. 4. Offer a hypothesis for each major move, clearly labelled as a hypothesis. 5. Flag any data gaps and remind me to verify all values against the originals. </task>
A multi-period ratio matrix with trend direction and hypotheses for the biggest moves.
Pro tip: Keep figures in a consistent unit (e.g. all in thousands) across periods so ChatGPT doesn't mix scales.
Prompts get you started. Tutorials level you up.
A growing library of 300+ hands-on AI tutorials. New tutorials added every week.
Statement & Trend Analysis
5 promptsCommon-Size Income Statement
6/30<context> Build a common-size income statement for [COMPANY]. Income statement figures: [FIGURES] (revenue and each expense/income line). Period: [PERIOD]. </context> <task> 1. Express every line item as a percentage of total revenue. 2. Output a table: Line item | Amount | % of revenue. 3. Identify the three largest cost lines as a share of revenue. 4. Comment on where margin is being lost or preserved. 5. State explicitly that you have not changed any reported figure, only re-expressed it, and ask me to verify the percentages. </task>
A percentage-of-revenue view that exposes cost structure and margin drivers.
Pro tip: Common-size statements make ChatGPT's output comparable across companies of different sizes — ask for it before any peer comparison.
Common-Size Balance Sheet
7/30<context> Build a common-size balance sheet for [COMPANY]. Balance sheet figures: [FIGURES] (every asset, liability, and equity line). Period: [PERIOD]. </context> <task> 1. Express each asset line as a % of total assets and each liability/equity line as a % of total liabilities + equity. 2. Output two tables (assets; liabilities & equity) with Amount and %. 3. Note the composition of the asset base (e.g. asset-heavy vs. asset-light) and the funding mix. 4. Flag any line that exceeds a reasonable concentration threshold. 5. Confirm totals foot to 100% and ask me to verify. </task>
Asset and funding composition expressed as percentages, with concentration flags.
Pro tip: Ask ChatGPT to confirm the percentages sum to 100% — a quick footing check that catches transcription errors.
Horizontal (Year-over-Year) Trend
8/30<context> Perform horizontal analysis for [COMPANY]. Multi-period figures: [FIGURES] (same line items across at least two periods). Periods: [PERIOD]. </context> <task> 1. For each line item, compute the absolute change and the % change versus the prior period. 2. Output a table: Line item | Prior | Current | Change | % change. 3. Rank the five line items with the largest % moves. 4. Group the moves into a short narrative (growth, cost pressure, one-offs) labelled as interpretation. 5. Remind me that % changes off small bases can mislead, and to verify the underlying numbers. </task>
Year-over-year absolute and percentage changes with the most material movers ranked.
Pro tip: Watch for large percentage swings on tiny base numbers — ask ChatGPT to flag any change where the prior value is near zero.
Cash Flow Statement Read
9/30<context> Interpret the cash flow statement for [COMPANY]. Figures: [FIGURES] (operating, investing, financing cash flows; net income; capex; D&A). Period: [PERIOD]. </context> <task> 1. Summarize cash generated/used by each activity and the net change in cash. 2. Compute free cash flow (operating cash flow - capex) and the cash flow to net income ratio. 3. Assess earnings quality: does operating cash flow support reported net income? 4. Flag any reliance on financing inflows to fund operations. 5. Note that this is analysis support, not advice, and ask me to verify the figures. </task>
A cash flow read covering FCF, earnings quality, and financing reliance.
Pro tip: A persistent gap between net income and operating cash flow is a quality flag — ask ChatGPT to quantify it explicitly.
Statement Linkage Check
10/30<context> Check consistency across the three statements for [COMPANY]. Figures: [FIGURES] (net income, depreciation, change in retained earnings, ending cash, beginning cash). Period: [PERIOD]. </context> <task> 1. Verify that net income flows correctly into retained earnings and the cash flow statement. 2. Confirm the cash flow statement's ending cash matches the balance sheet cash. 3. List each linkage checked with a Pass/Fail and the figures compared. 4. For any Fail, describe what could explain the discrepancy. 5. State clearly that a Pass here does not guarantee accuracy and that source data must still be verified. </task>
A linkage audit confirming the three statements tie together, with pass/fail per check.
Pro tip: Use this as a sanity gate before deeper analysis — if the statements don't tie, every downstream ratio is suspect.
Variance Analysis
5 promptsBudget vs. Actual Variance
11/30<context> Run budget-versus-actual variance analysis for [COMPANY]. Figures: [FIGURES] (budgeted and actual amounts per line item). Period: [PERIOD]. </context> <task> 1. For each line, compute the variance (actual - budget) and the % variance, labelling favorable (F) or unfavorable (U). 2. Output a table: Line | Budget | Actual | Variance | % | F/U. 3. Apply a materiality threshold of [THRESHOLD] and list only variances that breach it. 4. For each material variance, list plausible drivers to investigate (do not assert causes). 5. Remind me to verify the figures and confirm the favorable/unfavorable convention matches our reporting. </task>
A budget-vs-actual table with favorable/unfavorable flags and material variances isolated.
Pro tip: Set a materiality threshold up front so ChatGPT surfaces only the variances worth a manager's attention.
Price vs. Volume Decomposition
12/30<context> Decompose a revenue variance for [COMPANY] into price and volume effects. Figures: [FIGURES] (budget vs actual units and price/unit; or total revenue with unit and price detail). Period: [PERIOD]. </context> <task> 1. Compute the price variance and volume variance using the standard decomposition (price variance = (actual price - budget price) x actual volume; volume variance = (actual volume - budget volume) x budget price). 2. Show the formula, the inputs, and the result for each. 3. Confirm the two variances sum to the total revenue variance. 4. State which effect dominates and what it suggests about demand vs. pricing. 5. Ask me to verify the reconciliation and the input figures. </task>
A price/volume bridge that explains how much of a revenue variance came from each effect.
Pro tip: Always make ChatGPT reconcile price + volume variance back to the total — it confirms the decomposition is arithmetically sound.
Cost Variance Investigation
13/30<context> Investigate operating cost variances for [COMPANY]. Figures: [FIGURES] (standard/budget cost and actual cost per category, with volume if available). Period: [PERIOD]. </context> <task> 1. Compute the variance and % variance for each cost category. 2. Where volume data exists, split into rate and efficiency/usage components and show the formulas. 3. Rank categories by absolute unfavorable variance. 4. For the top three, propose specific questions to ask the cost owner — not assumed answers. 5. Note any category lacking enough data to split, and remind me to verify all figures. </task>
Cost variances split into rate and usage effects with investigation questions for owners.
Pro tip: Ask for the rate-vs-usage split only where you supplied volume data; otherwise ChatGPT may fabricate the components.
Forecast vs. Actual (Rolling)
14/30<context> Compare the latest forecast to actuals for [COMPANY]. Figures: [FIGURES] (prior forecast and actual results by line and by month/quarter). Period: [PERIOD]. </context> <task> 1. Compute forecast accuracy per line: variance, % variance, and absolute error. 2. Identify which lines are consistently over- or under-forecast across periods. 3. Summarize overall forecast bias (optimistic/pessimistic) with supporting numbers. 4. Recommend which forecast assumptions to revisit, framed as suggestions. 5. Remind me that conclusions depend on data accuracy and must be verified. </task>
A rolling forecast-accuracy review that exposes systematic forecasting bias.
Pro tip: Feeding several periods at once lets ChatGPT spot persistent bias that a single-period variance would hide.
Variance Commentary Draft
15/30<context> Draft management commentary on this period's variances for [COMPANY]. Figures: [FIGURES] (key variances and their confirmed drivers, supplied by me). Period: [PERIOD]. Audience: [AUDIENCE]. </context> <task> 1. Using only the drivers I provided, write 4-6 concise commentary bullets explaining the most material variances. 2. Lead each bullet with the variance figure, then the driver, then the implication. 3. Keep the tone factual and suited to [AUDIENCE]; avoid speculation beyond the drivers given. 4. Add a one-line outlook only if I supplied forward data; otherwise omit it. 5. End by noting the figures should be reconciled to the close before distribution. </task>
Ready-to-edit variance commentary bullets grounded only in drivers you supply.
Pro tip: Supply the confirmed drivers yourself — never let ChatGPT invent the "why" behind a variance for a financial report.
Forecasting & Scenarios
5 promptsDriver-Based Revenue Forecast
16/30<context> Build a driver-based revenue forecast for [COMPANY]. Figures: [FIGURES] (historical revenue, units, price, growth rates, and the drivers I want to use). Forecast horizon: [PERIOD]. </context> <task> 1. Lay out the forecast logic: revenue = driver 1 x driver 2 (e.g. customers x ARPU), and state each assumption explicitly. 2. Project revenue for each period in the horizon, showing the math. 3. Output a table: Period | Driver values | Forecast revenue. 4. List every assumption in one place so I can challenge each. 5. Caveat that this is a model, not a prediction, and that inputs must be verified. </task>
A transparent driver-based revenue model with assumptions isolated for review.
Pro tip: Ask ChatGPT to list assumptions separately from the math so you can swap a number without rebuilding the logic.
Three-Scenario Model
17/30<context> Build base, bull, and bear scenarios for [COMPANY]. Figures: [FIGURES] (base-case assumptions for revenue growth, margins, and key costs). Horizon: [PERIOD]. </context> <task> 1. Define the assumption changes for each scenario (e.g. bull = +X% growth, bear = -Y% growth) — ask me to confirm the deltas if I did not specify them. 2. Project revenue, operating income, and net income for each scenario across the horizon. 3. Output a table: Metric | Bear | Base | Bull per period. 4. Quantify the spread between bull and bear at the final period. 5. Note the key assumption that drives the widest spread and remind me to verify inputs. </task>
Bear/base/bull projections with the spread quantified and the dominant driver identified.
Pro tip: Pin down the exact percentage deltas for each scenario before generating — vague "optimistic/pessimistic" labels produce unreliable numbers.
Sensitivity Analysis
18/30<context> Run a sensitivity analysis for [COMPANY] on a target metric. Figures: [FIGURES] (base-case model, the output metric, and the inputs to flex). Target metric: [METRIC]. </context> <task> 1. Hold all inputs at base case, then flex one input at a time by -10%, -5%, +5%, +10%. 2. Show how [METRIC] responds to each input change in a table: Input | -10% | -5% | Base | +5% | +10%. 3. Rank inputs by how much they move [METRIC] (most to least sensitive). 4. State which input the result is most fragile to. 5. Caveat that flexing inputs independently ignores correlations, and ask me to verify the base model. </task>
A one-variable-at-a-time sensitivity table ranking which inputs matter most.
Pro tip: Single-variable sensitivity ignores that drivers move together — note this when one input dominates the result.
Cash Runway & Burn Projection
19/30<context> Project cash runway for [COMPANY]. Figures: [FIGURES] (current cash balance, monthly revenue, monthly operating costs, expected changes). Horizon: [PERIOD]. </context> <task> 1. Compute monthly net burn (or net cash generation) and the projected cash balance for each month. 2. Output a table: Month | Inflow | Outflow | Net | Ending cash. 3. State the month, if any, when cash reaches zero (runway). 4. Show how runway shifts if burn rises or falls by [DELTA]%. 5. Caveat that timing assumptions are critical and figures must be verified against the latest actuals. </task>
A month-by-month cash projection with runway and a burn-rate sensitivity.
Pro tip: Runway is highly sensitive to timing — ask ChatGPT to flag any month where an inflow assumption is doing most of the work.
Forecast Assumption Stress Test
20/30<context> Stress-test the assumptions behind [COMPANY]'s forecast. Figures: [FIGURES] (the forecast and its underlying assumptions). Period: [PERIOD]. </context> <task> 1. List every assumption embedded in the forecast that you can identify. 2. For each, classify the risk level (low/medium/high) and explain why. 3. Identify which two assumptions, if wrong, would most damage the forecast. 4. Propose a leading indicator to monitor for each high-risk assumption. 5. Remind me that this is analysis support and that I must validate the assumptions and figures. </task>
A risk-ranked inventory of forecast assumptions with leading indicators to watch.
Pro tip: Have ChatGPT surface the hidden assumptions you did not state explicitly — those are usually the riskiest.
Valuation
5 promptsComparable Multiples (Comps)
21/30<context> Value [COMPANY] using trading multiples. Figures: [FIGURES] (target metrics — revenue, EBITDA, earnings — and peer multiples I supply). Valuation date: [PERIOD]. </context> <task> 1. Apply each supplied peer multiple (EV/Revenue, EV/EBITDA, P/E) to the matching target metric. 2. Output a table: Multiple | Peer value | Target metric | Implied value. 3. Present an implied valuation range (low/median/high) across the multiples. 4. Note why each multiple might over- or under-state value for this company. 5. State that comps are only as good as the peer set and that all inputs must be verified — this is not investment advice. </task>
An implied valuation range from peer multiples with caveats on comparability.
Pro tip: Supply the peer multiples yourself from a current source — ChatGPT should not recall market multiples from memory.
Simple DCF Walkthrough
22/30<context> Walk through a simplified DCF for [COMPANY]. Figures: [FIGURES] (projected free cash flows, discount rate, terminal growth rate, net debt, shares outstanding). Horizon: [PERIOD]. </context> <task> 1. Discount each projected free cash flow to present value, showing the discount factor and math per period. 2. Compute the terminal value (Gordon growth) and its present value, showing the formula. 3. Sum to enterprise value, subtract net debt for equity value, and divide by shares for value per share. 4. Output each step in a clear table. 5. Caveat heavily that DCF output is highly sensitive to the discount and terminal-growth rates, and require me to verify every input. Not investment advice. </task>
A step-by-step DCF from cash flows to value per share with every intermediate shown.
Pro tip: Ask ChatGPT to show the discount factor for each period — it lets you spot a compounding error immediately.
WACC Build-Up
23/30<context> Estimate WACC for [COMPANY]. Figures: [FIGURES] (cost of equity inputs — risk-free rate, beta, equity risk premium; cost of debt; tax rate; market values of debt and equity). Period: [PERIOD]. </context> <task> 1. Compute cost of equity via CAPM (risk-free + beta x equity risk premium), showing inputs. 2. Compute after-tax cost of debt. 3. Weight by the capital structure to derive WACC, showing the weights and formula. 4. Output a table of every input and the final WACC. 5. State that small input changes move WACC materially, ask me to verify all inputs, and note this is not investment advice. </task>
A transparent WACC calculation with CAPM cost of equity and capital-structure weights.
Pro tip: Provide the risk-free rate and equity risk premium yourself — these are time-sensitive and should not come from model memory.
Asset-Based / Book Value
24/30<context> Estimate value for [COMPANY] on an asset basis. Figures: [FIGURES] (assets and liabilities at book; any fair-value adjustments I provide). Period: [PERIOD]. </context> <task> 1. Compute book equity (total assets - total liabilities). 2. Apply any fair-value adjustments I supplied and recompute adjusted net asset value. 3. Output a table: Item | Book | Adjustment | Adjusted. 4. Explain when an asset-based approach is most and least appropriate for this company. 5. Caveat that book values may diverge from market values and that figures must be verified. Not investment advice. </task>
A net-asset-value estimate with optional fair-value adjustments and applicability notes.
Pro tip: Asset-based valuation suits asset-heavy firms; ask ChatGPT to flag if it's a poor fit for a services or IP-driven business.
Valuation Cross-Check
25/30<context> Reconcile multiple valuation approaches for [COMPANY]. Figures: [FIGURES] (value-per-share or equity value from each method you have run: comps, DCF, asset-based). Period: [PERIOD]. </context> <task> 1. Tabulate the value from each method side by side. 2. Compute the range and the spread between the highest and lowest. 3. Explain the most likely reasons the methods disagree. 4. Suggest which method deserves the most weight for this company and why — as analysis, not a recommendation. 5. Caveat that no single method is definitive, require verification of all inputs, and state this is not investment advice. </task>
A side-by-side reconciliation of valuation methods with reasons for divergence.
Pro tip: When methods diverge widely, ask ChatGPT to trace the divergence to a specific assumption rather than averaging the outputs.
Go from copy-pasting to actually mastering AI.
AI Academy: 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ other tools. New tutorials added every week.
Reporting & Communication
5 promptsExecutive Financial Summary
26/30<context> Write an executive summary of [COMPANY]'s financial results. Figures: [FIGURES] (headline metrics, key variances, and confirmed drivers I supply). Period: [PERIOD]. Audience: [AUDIENCE]. </context> <task> 1. Open with a 2-3 sentence headline on overall performance, grounded only in the figures provided. 2. Add 4-5 bullets on revenue, margin, cash, and the most material variance with its confirmed driver. 3. Keep it under 200 words and tuned to [AUDIENCE]. 4. Avoid jargon and do not introduce numbers or drivers I did not provide. 5. End with a note that figures are subject to verification before distribution. </task>
A concise, audience-tuned executive summary built only from the data you provide.
Pro tip: Specify the audience (board, lender, team) — the same numbers need a very different framing for each.
Plain-Language Metric Explainer
27/30<context> Explain a financial metric for a non-finance audience at [COMPANY]. Metric and figures: [FIGURES] (the metric, its value, and how it was calculated). Audience: [AUDIENCE]. </context> <task> 1. Define the metric in one plain sentence with a relatable analogy. 2. Explain what this specific value means for [COMPANY], using the figure provided. 3. State what would make the metric better or worse. 4. Note one common misconception about the metric. 5. Keep it jargon-free and remind the reader the figure should be verified against the source. </task>
A jargon-free explanation of a metric and what its value means in context.
Pro tip: Great for prepping non-finance stakeholders before a review — ask for an analogy tied to their domain.
Board Slide Talking Points
28/30<context> Draft talking points for a board finance update for [COMPANY]. Figures: [FIGURES] (the metrics and confirmed narrative points I supply). Period: [PERIOD]. </context> <task> 1. Produce 5-7 crisp talking points, each anchored to a specific figure I provided. 2. Group them under: Performance, Cash & Liquidity, Risks, Outlook. 3. For each risk, pair it with the mitigation I supplied (do not invent mitigations). 4. Keep each point to one sentence a presenter can deliver verbally. 5. Flag any section where I have not provided enough data, and note figures need verification. </task>
Grouped, deliverable board talking points anchored to your figures.
Pro tip: Ask ChatGPT to flag thin sections rather than padding them — gaps you can fill beat invented content in a board setting.
KPI Dashboard Definition
29/30<context> Define a finance KPI dashboard for [COMPANY]. Context: [FIGURES] (the metrics I want to track and any current values). Audience: [AUDIENCE]. </context> <task> 1. For each KPI, specify: definition, formula, data source, target/benchmark, and reporting frequency. 2. Output a table with those columns. 3. Group KPIs into Growth, Profitability, Liquidity, and Efficiency. 4. Recommend which 3-4 KPIs belong "above the fold" for [AUDIENCE]. 5. Note that targets are placeholders to be confirmed and that values must be verified. </task>
A structured KPI dashboard spec with formulas, sources, and a prioritized headline set.
Pro tip: Name the data source for each KPI now — it forces clarity on whether the number is even reliably available.
Earnings/Results Q&A Prep
30/30<context> Prepare anticipated questions on [COMPANY]'s results. Figures: [FIGURES] (the results, key variances, and confirmed context I supply). Audience: [AUDIENCE] (e.g. investors, lenders, leadership). </context> <task> 1. Generate 8-10 tough questions [AUDIENCE] is likely to ask, based only on the figures and variances provided. 2. For each, draft a factual answer grounded in the supplied data; where I have not provided the answer, mark it "NEED DATA". 3. Order questions from most to least likely. 4. Flag any question that touches a sensitive or unresolved area. 5. Remind me to verify all figures and have answers reviewed before the meeting. </task>
A ranked Q&A prep list with grounded answers and clear gaps to fill.
Pro tip: The "NEED DATA" flags are the point — they tell you exactly what to nail down before you face the room.
Frequently Asked Questions
Prompts are the starting line. Tutorials are the finish.
A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.
7-day free trial. Cancel anytime.