Building a first financial model that isn't a fantasy

Most first financial models are wish lists dressed up as spreadsheets. They open with last year's revenue, add a confident "we'll grow 20%," let costs drift up by some smaller number, and bottom out at a profit that always seems to land right where the founder hoped. That model can't help you make a single real decision, because it never connected the numbers to anything you actually do.

A real model is different. It's a connected set of assumptions that drives your revenue, then your costs, then your cash — in that order, with each step traceable to the one before it. When you change one input, the whole thing moves, and you can see exactly why. That property is the entire point. A model that doesn't move when you change your mind isn't a model; it's a guess with formatting.

You don't need to be an accountant to build one that works. You need to be honest about what drives your business and disciplined about keeping your guesses visible. Here's how to do that.

Build revenue from drivers, not from hope

Top-down growth ("we did $2 million, next year we'll do $2.4 million") tells you nothing about whether $2.4 million is achievable. Bottom-up revenue starts from the things you can actually influence and multiplies them up. For a product business, that's units sold times price. For a subscription or services business, it's number of customers times average contract value times the share who stick around — your retention rate, meaning the percentage of customers still paying a year later.

Say you charge $500 a month and you add 8 new customers a month, but 3% of your base cancels each month. Now you have a model that can answer real questions: what happens if you lift price to $550, or pull churn down from 3% to 2%, or hire a second salesperson who adds 5 more customers a month? Each of those is a lever you can pull in the real world, and the spreadsheet shows you what it's worth before you commit a dollar to it.

Tie every cost to something that causes it

Costs aren't one blob. They split into three kinds, and each behaves differently as you grow. Treating them as a single line that "goes up a bit" is how models lie to you.

• Headcount-driven costs: build a simple hiring plan — who you hire and when — and let it generate salaries, payroll taxes, and benefits. People are usually your biggest line, so model them by name and start date, not as a lump sum.
• Variable costs: these move with revenue, so express them as a percentage of it. If payment processing and hosting run 9 cents on every revenue dollar, model them as 9% — they grow automatically as sales grow.
• Fixed costs: rent, software subscriptions, insurance. List these explicitly, line by line, because they stay flat until you make a decision to change them.

When costs are wired to their causes this way, growth has a price tag attached. Doubling customers might mean two more support hires and a bigger hosting bill — and your model says so, instead of pretending scale is free.

Keep your assumptions in one visible place

The single biggest upgrade to a homemade model is separating assumptions from calculations. Put every number you're guessing at — price, churn, hires, growth rate, cost percentages — in one labeled section at the top or on its own tab. Everywhere else, your formulas point back to those cells and never contain a typed-in number of their own.

Do this and flexing the model becomes a five-second job: change churn from 3% to 4% in the one place it lives, and watch revenue, headcount, and cash all respond. Bury that 3% inside forty different formulas and you'll never trust the output again, because you can't find every place it hides. Visible assumptions are what make a model honest — anyone can see what you believe and challenge it.

Use it to test, not to predict

A model's value isn't the final number; it's the conversation it lets you have with reality. The right question is never "what will happen?" It's "what if?" What if we grow half as fast? What if our biggest customer leaves? What if we delay that hire by a quarter — does cash hold? Run those, and you stop managing toward a fantasy and start preparing for the range of outcomes you might actually get.

This is also why you should distrust the hockey stick — the chart where revenue rockets up while costs stay suspiciously flat. Real growth costs real money: more people, more infrastructure, more support. If your model shows revenue tripling while expenses barely move, you haven't found a miracle; you've forgotten to link your costs to your drivers. Fix that, and the curve gets honest in a hurry.

Start small. Open a clean spreadsheet, put your real drivers in one assumptions block, and build revenue then costs then cash off of them — then break it on purpose with a few "what ifs." A rough model you can flex beats a polished one you can only admire.