Build Financial Models with AI

1. 1

   Define Model Scope and Key Assumptions

   Establish what your model needs to do and the critical assumptions that will drive every output. Bad assumptions produce confidently wrong models — this step prevents that.

   ChatGPT Claude

   I need to build a financial model for my business. Help me set up the foundations.
   
   **Business type**: [e.g., 'B2B SaaS startup,' 'e-commerce brand,' 'brick-and-mortar restaurant,' 'services business,' 'marketplace']
   **Purpose of the model**: [e.g., 'internal planning and cash flow management,' 'fundraising — investor deck,' 'bank loan application,' 'acquisition due diligence']
   **Time horizon**: [e.g., '3-year forecast' or '5-year forecast']
   **Current stage**: [e.g., 'pre-revenue,' '$50K MRR,' '$2M ARR']
   **Key business metrics I track**: [e.g., MRR, churn rate, CAC, LTV, gross margin, headcount]
   
   1. **Model Architecture Recommendation**: Given my business type and purpose, what should this model include? Recommend the key tabs/sections:
      - Revenue model (and which revenue modeling approach fits my business type)
      - Cost structure (COGS vs. OpEx, key cost line items for my business type)
      - Headcount and payroll
      - Cash flow statement
      - P&L summary
      - Balance sheet (necessary for bank applications; optional for internal use)
      - Key metrics dashboard
      - Scenario analysis
   
   2. **Critical Assumptions Identification**: For my business type, what are the 8-12 assumptions that have the most leverage over the model's outputs? These are the inputs I should spend the most time validating, not estimating. For each assumption, tell me: (a) what it is, (b) what a reasonable range looks like for a business at my stage, (c) where I can find benchmark data.
   
   3. **Revenue Model Design**: What is the right revenue modeling structure for my business? Design the logic: what are the key drivers (e.g., for SaaS: new logos + expansion - churn; for e-commerce: sessions × conversion rate × AOV × repeat rate). Write out the formula logic before I build it.
   
   4. **Data I Need to Gather**: What historical data or market benchmarks do I need to pull together before I can populate this model? Give me a specific list.

   Copy

   Tip: A financial model is only as good as its critical assumptions. Spend 80% of your model-building time validating the 3-4 assumptions with the most leverage over your outputs. For most startups: growth rate, churn rate, and gross margin. For most businesses: customer acquisition cost and retention. Get those right; the model will be useful.

2. 2

   Build the Revenue Model

   Design the revenue forecasting section — the core of most financial models — with the logic, drivers, and formulas specific to your business model.

   ChatGPT Claude

   Help me design and build the revenue section of my financial model.
   
   **Business model**: [describe your revenue model — subscription, transactional, project-based, etc.]
   **Revenue streams**: [list all ways you make money]
   **Key metrics I know or can estimate**: [e.g., current MRR, growth rate, churn rate, average deal size, sales cycle length]
   
   For EACH revenue stream, design the model logic:
   
   1. **Revenue Driver Tree**: Write out the formula logic as a driver tree. Example for SaaS MRR:
      - End-of-month customers = beginning customers + new customers - churned customers
      - New customers = leads × qualified rate × close rate
      - Churned customers = beginning customers × monthly churn rate
      - MRR = end-of-month customers × ARPU
      - Then expand to include expansion MRR and contraction MRR
   
   2. **Spreadsheet Structure**: Design the exact row-by-row structure I should use in my spreadsheet. Give me:
      - Assumption inputs (hardcoded, clearly labeled, in one section)
      - Calculation rows (formulas that reference assumption inputs only, never hardcoded numbers in formulas)
      - Output/summary rows
      Tell me which rows should be monthly granularity and which annual.
   
   3. **Growth Assumptions**: For my stage and business type, what are realistic growth assumptions for [Year 1 / Year 2 / Year 3]? Give me benchmarks: what growth rates are (a) conservative, (b) base case, (c) optimistic for a company at my stage? Source these benchmarks from industry data where possible.
   
   4. **Seasonality**: Does my business type typically have seasonality? If so, how should I model it? Give me a monthly distribution factor table I can apply.
   
   5. **Excel/Sheets Formula Examples**: Write the actual formulas for the 5 most complex calculations in my revenue model, in Excel syntax. Include comments explaining the logic.

   Copy

   Tip: In your model, separate assumptions (the numbers you're guessing) from calculations (the formulas that process those numbers). All assumptions should live in one clearly labeled section with a different background color. Never hardcode a number into a formula that calculates something else. This separation makes scenarios trivial to run and errors easy to find.

3. 3

   Build the Cost Model and P&L

   Design the cost structure, headcount plan, and full P&L that connects revenue to the bottom line.

   ChatGPT Claude

   Help me design the cost model and P&L for my [business type] financial model.
   
   **My business**: [description]
   **Current team size**: [number and roles]
   **Key cost categories I currently have**: [e.g., 'AWS hosting, Stripe fees, salaries for 4 people, office rent']
   **Revenue assumptions** (from Step 2): [paste revenue model summary]
   
   1. **Cost of Goods Sold (COGS)**:
      - For my business type, what belongs in COGS vs. OpEx? Give me the correct classification.
      - Design the COGS model: which COGS items are variable (% of revenue) and which are fixed?
      - What gross margin should I target for my business type? What are the industry benchmarks (seed-stage vs. mature)?
      - Write out the COGS driver tree (like the revenue driver tree in Step 2)
   
   2. **Headcount Plan**:
      - For my stage and growth plan, what hiring sequence is typical? When do you typically hire your first X role?
      - Design a headcount model: list each role, the month they're hired, their fully-loaded cost (salary + benefits + payroll taxes — typically 1.2-1.3x salary), and which cost category they belong in (COGS, Sales & Marketing, R&D, G&A)
      - How does headcount typically scale relative to revenue for my business type?
   
   3. **Operating Expense Structure**:
      - List the standard OpEx categories for my business type
      - For each category, should it be modeled as: fixed, variable (% of revenue), or headcount-driven?
      - Suggest reasonable expense ratios as a % of revenue for each category at my stage
   
   4. **P&L Structure**:
      Design the complete P&L layout, from revenue to net income:
      Revenue → Gross Profit → [OpEx categories] → EBITDA → [D&A and interest] → Net Income
      Include which metrics to calculate as % of revenue on each line for easy benchmarking.

   Copy

   Tip: Headcount is almost always the dominant cost for service and software businesses — model it carefully and specifically. Don't model 'headcount grows proportionally with revenue.' Instead, model specific hires in specific months with specific roles and costs. The headcount schedule is often where unrealistic models become obvious — you'll see that the business you're projecting can't physically be run by the team you've planned.

4. 4

   Build Scenarios and Sensitivity Analysis

   Design the scenario analysis and sensitivity tables that show how your model performs under different conditions — turning a single-point forecast into a decision-making tool.

   Claude ChatGPT

   Help me build the scenario and sensitivity analysis for my financial model.
   
   **Model summary**: [paste key outputs — revenue projections, burn rate, path to profitability from Steps 2-3]
   **My most uncertain assumptions**: [the 3-5 assumptions you're least confident about]
   **Model purpose**: [fundraising / planning / bank application]
   
   1. **Three Scenarios**: Design a Base, Bear (downside), and Bull (upside) scenario. For each:
      - Change only the most impactful 3-5 assumptions — not everything
      - Describe what real-world situation each scenario represents (e.g., Bull = product-market fit accelerates; Bear = we don't achieve PMF in Year 1)
      - Show the key output differences: Year 3 revenue, runway, path to profitability (or lack thereof)
      - For investor use: which scenario should I present as 'the plan'? (hint: not the bull case)
   
   2. **Sensitivity Table Design**: Build two sensitivity tables:
      - Table 1: Net income or EBITDA margin at Year 3 as a function of [growth rate] × [gross margin]
      - Table 2: Cash runway (months) as a function of [monthly burn rate] × [monthly revenue growth rate]
      Design these as data tables with the assumption ranges across axes. Specify the exact range of inputs for each axis.
   
   3. **Break-Even Analysis**:
      - At current burn rate and growth rate, when does the business break even on a monthly basis?
      - What growth rate is required to break even by [target date]?
      - What is the minimum MRR/revenue required to sustain the current team at current headcount?
   
   4. **Investor Narrative**: Given the model outputs, write the 3-sentence financial narrative for an investor meeting: what the numbers show, what the key risks are, and what achieving the plan requires. Be honest about risks — investors respect it more than false confidence.
   
   5. **Model Audit Checklist**: Before I present this model to anyone, what should I check? Write a 10-point checklist for model quality and accuracy.

   Copy

   Tip: Present your base case to investors, not your bull case. Sophisticated investors will back-calculate your assumptions and immediately notice if your base case is optimistic. When they see you've been conservative and clear-eyed about risks, it builds credibility. Being wrong with a well-reasoned model is fine; being caught presenting an inflated scenario is not recoverable.

Recommended Tools for This Scenario

ChatGPT

Freemium

The AI assistant that started the generative AI revolution

• GPT-4o multimodal model with text, vision, and audio
• DALL-E 3 image generation
• Code Interpreter for data analysis and visualization

Get Started →

Claude

Freemium

Anthropic's AI assistant built for thoughtful analysis and safe, nuanced conversations

• 200K token context window for massive document processing
• Artifacts — interactive side-panel for code, docs, and visualizations
• Projects with persistent context and custom instructions

Get Started →