A startup financial model that survives diligence is built from the bottom up: you count the people you plan to hire, price them at real market rates, and attach one defensible revenue assumption. Most founder models fail not because the math is wrong but because the inputs are invented — top-line growth pasted in at 15% a month with nothing underneath it. Your model should answer three questions on one screen: when does the money run out, what does the money buy, and what has to be true for the next round to close.
This is a practical drafting guide, not financial advice — your accountant and counsel should review anything you show to investors.
What should the core structure of the model be?
Use three linked sheets: an assumptions tab, a monthly operating model for 18–24 months, and a summary dashboard. Every number in the operating model should trace back to a cell on the assumptions tab — hire dates, fully loaded salaries, cloud unit costs, price per seat. The discipline matters because investors test models by changing one assumption and watching what moves. If a single input ripples cleanly through hiring, burn, and runway, the model reads as credible before anyone checks the arithmetic. Keep formulas simple enough that a person, not just a spreadsheet, can explain them.
How do you forecast revenue without lying to yourself?
Forecast revenue from units, not percentages. For a SaaS product that means seats × price × win rate × pipeline; for services it means billable people × utilization × rate. Per the U.S. Small Business Administration's business planning guidance, realistic projections tied to specific drivers are the foundation lenders and investors expect. Build three cases — base, upside, downside — and make the downside the one you manage against. A useful honesty test: your base case revenue for month 18 should rarely exceed 4–6× your current monthly revenue unless you can name the mechanism, such as a sales hire with a quota, that produces each step of the increase.
What does a realistic cost model include?
Start with headcount, because for most early-stage companies people are 60–70% of total spend. List every planned hire by month, at salary plus a 15–25% load for payroll taxes, benefits, and software per person. Then add the costs founders routinely forget: cloud bills that scale with usage, payment processing fees, insurance, legal retainer work around the next fundraise, and recruiting fees that can run 20–25% of first-year salary per hire. The U.S. Bureau of Labor Statistics publishes employer cost data you can use to sanity-check your load factor. Then stop. A cost model with forty line items nobody maintains is worse than twelve lines someone updates monthly.
How do you calculate runway honestly?
Runway equals cash balance divided by net burn, where net burn is cash out minus cash in — not accounting losses. Two adjustments make the number honest. First, use the burn rate from your last three actual months, not your budget, because budgets run optimistic. Second, model a hiring slip: assume your planned hires land one to two months late, which understates burn slightly but matches how recruiting actually goes. Recompute runway monthly and put the number at the top of the dashboard. When investors ask how long the money lasts, the founder who says "19 months at current burn, 16 if we make all hires on schedule" sounds like someone who runs a company.
What will investors test first?
Experienced investors usually probe the hiring plan and the gross margin before touching revenue. They ask why a hire lands in March rather than June, or why cloud costs stay flat while revenue triples — and "we'll optimize later" is the wrong answer in a diligence meeting. Have the linkage ready: this hire owns this metric, this metric drives this revenue line, this cost scales with this volume. Per reporting on 2025 venture trends by Crunchbase News, U.S. startups raised $328 billion in 2025, but the money concentrated in companies that could show this kind of operational specificity. The model is a story about cause and effect that happens to be written in cells.
How often should you rebuild versus update?
Update monthly in about an hour: actuals in, variance against plan noted, assumptions adjusted only when the world changed rather than when you missed the number. Rebuild the structure roughly once a year, or when the business model changes — a new pricing scheme or a shift from services to product makes the old model's plumbing obsolete. A monthly update ritual that survives is worth more than a gorgeous quarterly rebuild nobody has time for. Keep a dated copy of each month's model so you can see how your own forecasts drifted; the drift is the calibration lesson.
What are the most common modeling mistakes?
Five account for most of the broken models investors see: revenue that grows by assumption instead of mechanism; gross margin shown at 100% because infrastructure costs sit "below the line"; hiring plans with no load factor; founder salaries of zero, which signals the model is theater; and a runway number computed off budgeted burn instead of actual burn. Each mistake is easy to fix once named, and each fix makes the next conversation with an investor shorter.
- Build bottom-up: people, units, prices — never percentage growth pasted on top.
- Manage against the downside case; present the base case.
- Net burn from actuals, recomputed monthly, top of the dashboard.
Build the smallest model that answers the three questions, update it monthly, and let it embarrass you quietly in private so it never embarrasses you in a board meeting. Then stop adding tabs.
For more context, read Startup Runway Calculation: The Number That Decides Everything.
For more context, read open source business model.
For more context, read hire first startup engineers.