Guide

How to build a financial model for a startup: three statements, three years, one assumptions tab

A seed model built in the order it should be built, from the assumptions tab to a balance check of 0.00, with every step worked in real numbers on one sample company.

By Amer Habeeb. Updated . 8-minute read.

Every number below comes from Fernbrook Analytics, a fictional company. You can change its price in the live model on the home page.

What a seed model must answer: the six questions

A seed financial model is not a forecast you are graded on. It turns a few plain assumptions into three statements, so that an investor can change one input and see what happens to the cash.

The model has to answer six questions, the same six I check on every deck: runway, revenue drivers, unit economics, headcount, use of funds and tie-out. If the model cannot answer one of them, the deck built on it cannot either.

This guide builds that model in order, on Fernbrook Analytics, a fictional company that sells daily inventory forecasts to restaurant groups. Its sample seed deck is built from this model, and you can follow every step in any spreadsheet. If you would rather hand the build to someone, the questions to ask a startup financial model consultant are these same six.

The assumptions tab first

Every input lives on one tab, and nowhere else. The statements hold only formulas that point back to it. That rule keeps the model honest: when an investor asks “what if you charge less?”, you change one cell and every statement moves with it.

Keep it short: each line a number you can defend in one sentence.

Revenue build: customers × price × 12

Revenue is never typed in. It is customers × price × 12 months, year by year, so that an investor can argue with the customer count or the price, both of which you can defend, instead of with a total, which you cannot.

If your revenue line grows by a fixed percentage, replace it with customers and price. Growth is something the plan produces, not something you assume.

Costs from headcount up

At seed, people are most of the cost. Build operating costs from a hiring plan, not from a total that looks affordable: heads × average salary, plus payroll tax, plus sales and marketing, plus software, rent and admin. When the four lines add up to the operating cost on the assumptions tab, the team slide and the financials slide cannot disagree.

This is the check that failed in the original Fernbrook deck. Its team slide showed 12 people in Year 2 at $90,000, against $900,000 of Year 2 operating costs. The headcount check shows why that gap is the first thing an investor finds.

Each total equals the operating cost on the assumptions tab, to the dollar. Keep the build on its own tab, so there is only ever one number.

Income statement: from revenue to net loss

The income statement is the shortest of the three. Revenue, less the cost of serving customers, gives gross profit. Gross profit, less operating costs, gives net income. With no debt and no tax, there is nothing else to add.

A loss in every year is normal at seed. What an investor reads is the direction: the Year 3 loss is smaller than Year 2, even as the team grows.

Cash flow: where the money actually goes

Net income is not cash. A growing company bills customers before it collects, so some revenue sits in receivables at every year end. The cash flow statement starts from net income, takes off the increase in receivables, and adds the equity raised.

Runway comes from this statement, never from the income statement. Divide the cash used in a year by 12 and you have the monthly burn the ask slide should quote.

Year 3 ends with $248,800, so the raise funds the whole plan. If you would like this runway check, and the other five, run on your own deck, send me the deck and a link to the model. You get one written page back, and every point names the slide, the number and the fix.

Balance sheet and the check

The balance sheet is where the three statements prove they describe the same company. Fernbrook owns two things, cash and receivables. It owes nothing, so everything it owns belongs to its shareholders: the equity raised, less the losses to date.

Add one row at the bottom: total assets minus total liabilities and equity. Fernbrook owes nothing, so here it is assets minus equity. That row is the balance check. It should read 0.00 in every year, at every input you can type.

When your check is not zero, do not plug the difference. Find the cell that breaks it. The usual cause is the one in the next section.

The typed-over cell: a model that balances by accident

The most common modeling mistake is a formula replaced by a typed number. It usually happens late, to make one total look right, and then it is invisible: the check still reads 0.00. It breaks only when an input changes, which is exactly what an investor does in a meeting.

You can watch this happen: open the sample model on the home page, switch on the typed-over cell and move the price. To find typed cells in your own model, turn on the formula view in Excel or Google Sheets and look for a plain number in a row of formulas.

Price sensitivity: move one input, watch the cash

Once the model is built from one tab, sensitivity costs nothing: change one input and read Year 3 cash. Start with price, the assumption an investor is most likely to push on.

Put the answer on the slide before anyone asks: the price at which the plan stops being funded, and why you expect to stay above it.

The files to keep, and what to send with your deck

Keep the model as a live spreadsheet, in Excel or Google Sheets, with formulas intact. Keep four things together:

  • The assumptions tab, with a one-line source note for each input.
  • The cost build, with the hiring plan by year.
  • The three statements, every cell a formula.
  • The balance check row, visible on the first screen.

Keep a view-only link to the model ready, and send it with the deck when an investor asks for the numbers. Every figure on the deck should be one an investor can find in that file. If you have no model yet, the deck plus model package is $295, delivered in 7 days on Fiverr.

Check yours

  1. Every input sits on one assumptions tab, and the statements hold only formulas.
  2. Revenue is customers × price × 12 in every year, with no typed totals.
  3. Heads × salary, plus payroll tax and the other cost lines, equals operating costs in every year.
  4. Gross profit and net income follow from revenue and costs, with nothing typed in between.
  5. Cash flow starts from net income, takes off new receivables and adds the equity raised.
  6. The monthly burn on the deck comes from the cash flow statement.
  7. The balance check reads 0.00 in every year.
  8. Change the price, read the check again, and note the price at which cash runs out.

Send the deck you have, and the model link if there is one. Within 2 business days (India time) you get one written page: where investors will push hardest, two fixes you can make today, and what already holds up. It is a consistency check, not an audit.

Get my free Investor Read

Want the model built instead? See what the model build includes.

How this guide was made: written with Claude from the method I use on every read and the Fernbrook sample model. The self-test recomputes every Fernbrook number on this page; other figures link to their source.

Updated

Get my free Investor Read