Key points
- A diligence-ready model separates assumptions, calculations and outputs, so an analyst can change one driver and watch everything update.
- Revenue should be built bottom-up from drivers you can evidence: pipeline, conversion, pricing, churn and expansion.
- Investors test the model against your history. The first 12 to 24 months of actuals should match the accounting records.
- Show at least two scenarios: a base case and a slower case that still reaches the next milestone.
- Share an editable spreadsheet in the data room, view-only for most investors and downloadable for the lead's analyst.
A startup financial model for investors is a spreadsheet that projects revenue, costs, cash and headcount, usually for three to five years, from a set of stated assumptions. In due diligence its job is not to predict the future. Its job is to show that the founders understand what drives the business and what the money will buy.
Analysts at venture funds rebuild or stress-test models all the time. They look for the same things, and they find the same mistakes.
How investors use the model
| What the investor does | What they are checking | What helps |
|---|---|---|
| Reconciles actuals to the P&L | Is the model grounded in real numbers? | Actuals tab linked to accounting exports |
| Changes the growth driver | Does the plan rely on one heroic assumption? | All drivers on one assumptions tab |
| Looks at cash by month | When does the money run out? | Monthly cash flow, not just annual |
| Checks hiring against revenue | Is headcount growing faster than sales? | Headcount plan by role and start month |
| Compares to benchmarks | Are margins and efficiency plausible? | Gross margin, burn multiple and payback outputs |
| Runs a downside case | Can the company reach the next round if growth is slower? | Scenario switch on the assumptions tab |
Structure: three layers
Structure of a diligence-ready model
- 01Inputs
Every assumption in one place: pricing, conversion, churn, hiring dates, salaries, vendor costs.
- 02Actuals
Monthly history exported from the accounting system, unchanged.
- 03Calculations
Revenue build, cost build, headcount, cash. Formulas only, no typed numbers.
- 04Outputs
P&L, cash flow, balance sheet summary, key metrics, scenario comparison.
The rule that matters most: no hard-coded numbers in calculation cells. An analyst who finds a typed number in the middle of a formula chain stops trusting the rest of the model.
Revenue: build it from drivers
Top-down revenue (“we will take 1% of a $10 billion market”) tells an investor nothing. A bottom-up build does.
| Business model | Core drivers to model | Evidence for the data room |
|---|---|---|
| B2B SaaS, sales-led | Pipeline, win rate, sales cycle, ACV, ramp per rep, churn, expansion | CRM export, closed-won history, cohort table |
| B2B SaaS, self-serve | Traffic, signup rate, trial-to-paid, ARPA, churn | Product analytics, billing exports |
| Marketplace | Buyers, sellers, transactions per user, take rate | GMV history, cohort repeat rates |
| Consumer subscription | Installs, conversion, price, monthly churn | App store and billing data |
| Hardware | Units, price, bill of materials, channel margin | Purchase orders, supplier quotes |
For SaaS, the SaaS metrics guide defines each metric so your model and your metrics workbook agree.
Costs and headcount
Most startup costs are people. Model headcount by role with a start month and a fully loaded cost (salary, payroll taxes, benefits, equipment). Then add the costs that scale with revenue (hosting, payment fees, customer support) and the fixed costs (rent, software, insurance, legal).
Outputs investors read first
Use the same period for net burn and net new ARR. Below 1.5 is generally read as efficient for an early-stage company.
Scenarios
At minimum, show a base case and a slower case. In the slower case, growth and hiring both slow; the question is whether the company still reaches the milestones that justify the next round before cash runs out. Some founders add an upside case; investors will discount it, but it shows what more capital could do.
To test it, enter the cash after the round and the average monthly net burn from each case in the runway calculator.
Checks analysts run
- Actual revenue by month matches the accounting system.
- Opening cash matches the latest bank statement.
- Sum of monthly figures equals the annual totals.
- Balance sheet balances, if one is included.
- Headcount costs reconcile to payroll for the actuals period.
- Growth rate in month one of the forecast is consistent with the last three months of actuals.
- Churn and expansion assumptions are consistent with the cohort data.
- No circular references or broken links to other files.
Building and sharing the model
- Export actualsPull 12 to 24 months of monthly P&L and balance sheet data from your accounting system into an Actuals tab.
- Write down every assumptionPut each driver on an Inputs tab with a short note on where it comes from.
- Build revenue bottom-upModel customers, pricing, churn and expansion from the drivers, month by month.
- Build costs and headcountList hires by role and start month; add variable and fixed costs.
- Add scenariosA switch on the Inputs tab that changes growth, hiring and spend together.
- Run the checksReconcile actuals, check totals and remove hard-coded numbers.
- Share in the data roomUpload the working spreadsheet. Keep it view-only for most investors, allow download for the lead's analyst, and note the version date in the file name.
Version control matters. Investors who download a model on Monday and get a revised one on Thursday will want to know what changed. Keep a short change log tab and replace files in place so the data room shows the current version. If you need a room that handles this cleanly, the best data rooms for investors ranking compares the options.
How many years should a startup financial model cover?
Three to five years, with the first two years monthly. Investors pay most attention to the next 18 to 24 months.
Should the model be in Excel or Google Sheets?
Either is fine. What matters is that it is editable, formula-driven and opens without broken links.
Do investors expect the forecast to be accurate?
No. They expect it to be reasoned. They want to see which assumptions matter and how you would respond if they turn out wrong.
Should I share the model as a PDF?
Share the spreadsheet. A PDF hides the logic, and analysts will ask for the working file anyway.
What is the most common model mistake?
Hard-coded numbers inside formulas, and actuals that do not match the accounting records.
Do I need a balance sheet in the model?
At Series A and beyond, a summary balance sheet helps show working capital and cash. At pre-seed, a P&L and cash flow are usually enough.
