To build a financial model for fundraising, start from the drivers of your business rather than a target valuation. Build revenue bottom up, model headcount and burn without flattering the numbers, show the unit economics working, and back-solve the raise from the runway it buys. Investors read the model as proof of how you think. They already know the forecast itself is a guess.
That last point is the one most founders miss. You spend a weekend making the year-five number huge, and the person across the table has already stopped looking at it. What they are doing instead is quieter and more useful to understand.
What is a financial model for fundraising actually for?
A fundraising model is a document that lays out how your business turns effort and money into revenue over time. It usually covers three to five years, with the near term built month by month and the later years sketched at a higher level. It shows revenue, costs, headcount, cash burn, and how long your money lasts.
Here is the part that changes how you build it. Investors do not treat the model as a prediction. Paul Gompers of Harvard Business School and his co-authors surveyed 885 institutional venture capitalists and found that when they decide to invest. A smaller group, 17% of early-stage investors, use no quantitative deal metric whatsoever. These are not lazy investors. They have simply learned that a pre-revenue five-year forecast is a story dressed as math.
So if a third of them are not even running your numbers to a conclusion, why build the model at all? Because the model is proof of something the pitch cannot show on its own: whether you understand the machine you are building. The assumptions are the message. The output is just where the assumptions land.
Why the five-year hockey stick fails investors
Every curated document has the same problem. A deck shows your best slides. A reference list names people who like you. A five-year projection shows the future where everything works. Investors know this, so they discount the polish and go hunting for the logic underneath.
The hockey stick fails because it hides the two questions that actually decide the round. Can this business make money on each customer, and does the founder know it yet? CB Insights, in its post-mortem study of , found that 70% ran out of capital. That figure sounds like the cause of death. Running out of cash is the final symptom, and the disease sits upstream of it. Underneath the 70%, 43% cited poor product-market fit and 19% cited unsustainable unit economics. A model that projects a straight line to a hundred million dollars, while quietly assuming a customer costs nothing to acquire and never leaves, is describing a company that becomes one of that 19%.
This is why the model is a thinking artifact. When you build revenue from real drivers, the number of leads, the conversion rate, the price, the churn, you expose your understanding of the business to daylight. When you type a growth percentage into a cell and drag it across five years, you expose the opposite. Investors can tell the difference in about ninety seconds.
What investors actually do when they open your model
Picture the person opening your spreadsheet. They are not reading it left to right. They click into a few cells to see whether numbers are hard-typed or calculated. They pull one assumption to an extreme to see if the model breaks or bends. Then they check three chains.
The first chain is revenue to drivers. If revenue triples in year two, they want to see which driver tripled and whether that driver is believable. The second chain is spend to unit economics. Here they look at your burn multiple, which is the cash you burn for every dollar of new recurring revenue you add. Scale Venture Partners, using its Scale Studio dataset of several hundred SaaS companies, benchmarks the average burn multiple at , improving toward roughly 1.6x across the full dataset and 1.4x by the twenty-five to fifty million ARR band. A credible model shows that ratio getting better as you scale, because early companies are structurally inefficient and everyone knows it. A model that shows perfect efficiency at seed is a tell that the numbers were reverse-engineered from a happy ending.
The third chain is raise to runway to milestone. They divide your ask by your monthly burn to see how many months it buys, then ask what that runway is supposed to prove by the time it ends. If the answer is vague, the model has not done its job. If you want to see how investors think about the clock, our guide on covers why you should fund past the next raise, not up to it.
What investors look for in a financial model
Most of a fundraising model is scaffolding. A few parts carry the argument. A fundraising financial model contains six core parts: an assumptions tab, a bottom-up revenue build, a headcount and payroll schedule, operating costs that resolve into burn and runway, unit economics, and the raise back-solved from dilution. Build those six well and the rest can be simple.
Revenue, built from the bottom up. Start with the smallest real unit of your business and multiply up: this many outbound touches, this reply rate, this close rate, this price, this expansion. The market-share shortcut, "we capture 1% of a huge market," proves nothing, and investors know it. Bottom-up revenue is slower to build and far harder to argue with, which is exactly why they trust it.
Headcount and payroll. People are the biggest line item in almost every seed model, so this is where honesty shows. Carta's State of Seed 2025 report found that . Model your team off today's lean norm. A 2021 org chart is the wrong template. A hiring plan that adds twenty people in eighteen months signals that you learned to build during the last boom and have not adjusted.
Burn and runway. Sum your monthly costs, subtract revenue, and you have net burn. Divide cash by net burn and you have runway in months. This is the single most-checked output in the whole model, so it should be a clean calculation, not a typed-in figure.
Unit economics. Show what it costs to acquire a customer, called your CAC, and what that customer is worth over their life, called their LTV. If LTV is not clearly bigger than CAC, and heading in the right direction, the rest of the model does not matter. This is the same discipline investors bring to your live numbers, which our piece on breaks down in detail.
The ask, back-solved from dilution. Founders have consistently sold , per Carta. That norm lets you check your own raise. Take the amount your model says you need for eighteen to twenty-four months of runway, divide by roughly 0.20, and you get the post-money valuation you are implicitly asking for. If that number is wildly above what your stage supports, either the raise is too big or the plan is too expensive. Our guide on walks through sizing the ask against the milestones it needs to buy.
Here is the contrast that separates a model investors trust from one they discount.
Dimension | Projection-first model | Driver-based model |
|---|---|---|
Revenue | Growth percentage typed across years | Built from leads, conversion, price, churn |
Assumptions | Buried, hard to find | Listed on one tab, each one editable |
Headcount | Aggressive hiring plan | Tied to revenue milestones and today's lean norms |
Burn and runway | Stated as a comfortable number | Calculated from the cost build |
Unit economics | Implied, rarely shown | Explicit, and improving over time |
The raise | Set first, model bent to fit | Back-solved from runway and dilution |
What it proves | That you can use a spreadsheet | That you understand your business |
Why the bar moved up a full stage
The market context matters, because it changed what a "good" model has to show. PitchBook and the National Venture Capital Association reported that US venture firms raised . When the capital pool shrinks that far, investors fund fewer companies and demand more evidence per dollar.
That shows up directly in expectations. Waveup, in a , found that round expectations have jumped a full stage. Pre-seed founders are now expected to show what used to be seed-level traction, and seed rounds are held to standards that resembled Series A a few years ago. Net revenue retention, or NRR, the share of recurring revenue you keep and grow from existing customers, has become a metric investors ask about early. In this climate, a seed model has to lay out a believable path to Series A milestones, which for B2B SaaS now looks like roughly $3 million in ARR, a burn multiple under 1, and net revenue retention above 120%. This year's numbers alone no longer clear the bar. Your model is the place you prove that path is more than a wish.
Where a model stops being a spreadsheet and becomes proof
A spreadsheet makes a claim. Proof is when the claim is backed by something an investor cannot easily dismiss. The gap between the two is where most founders lose the room, because a model built entirely from assumptions has no anchor in reality yet.
This is the problem SeedForge was built to close. A founder runs one 30-minute AI session and connects real data sources, so the traction and revenue numbers behind the model come from live systems rather than a founder's optimism. The result is a Living Profile at seedforge.com, a single shareable link where an investor can see the model's assumptions sitting next to the real numbers that support them. The profile does not sit in a folder waiting for your next raise. It stays live and updates as you build, SeedForge runs matched outreach to relevant investors on a pay-per-outcome basis, and investor agents can keep watching the profile as your metrics move. You build the proof once, and it keeps working while you get back to the business. That is the difference between a model that argues for you in one meeting and proof that keeps arguing for you between them.
How to build the model, step by step
Open one tab called Assumptions and put every driver there. Nothing hard-typed anywhere else in the model should be a number that belongs on this tab.
Build revenue bottom up from those drivers. Resist the market-share shortcut.
Build the cost side, with headcount as its own schedule tied to revenue milestones.
Let burn and runway calculate themselves from the two sides. Never type the runway number.
Add a unit economics block: cost to acquire a customer, lifetime value, and how both move over time.
Back-solve the raise from the runway you need and the roughly 20% dilution norm, then sanity-check the implied valuation against your stage.
Build the near term month by month for at least eighteen months, and the outer years annually. Depth where it matters, sketch where it does not.
Stress one assumption to a bad case and keep that scenario in the file. Investors respect a founder who has already looked over the cliff.
Do this and the model becomes what it is supposed to be. A clear, honest account of how you turn money into a business, told in a way that survives someone pulling on the threads.
Frequently asked questions
How many years should a startup financial model project for fundraising?
Three to five years is standard, but the horizon matters less than the resolution. Build the first eighteen to twenty-four months month by month, since that is the runway your raise actually buys, and sketch the outer years annually. Investors scrutinize the near term and treat the far years as a directional story.
Do investors actually read startup financial models?
Some do closely and some barely glance. Harvard research on 885 venture capitalists found 31% of early-stage investors do not forecast cash flows at all. Even those who skip the math still open the model to test your assumptions and unit economics, so it is read as proof of your thinking rather than as a forecast they trust.
What is a burn multiple and what is a good one?
Burn multiple is the cash you burn for every dollar of new recurring revenue you add. Scale Venture Partners benchmarks the average at 3.4x for companies under one million in ARR, improving toward 1.6x as you scale. Early companies are expected to be inefficient, so investors want the ratio trending down. A perfect number at seed reads as reverse-engineered.
How much should I raise, and how do I show it in the model?
Size the raise off the runway it buys, aiming for eighteen to twenty-four months to a clear milestone. Carta data shows founders sell around 20% in a seed round, so dividing your target raise by roughly 0.20 gives the post-money you are implicitly asking for. If that valuation is far above your stage, the ask is too big.
Should I use a top-down or bottom-up revenue model?
Bottom-up, almost always. Top-down starts from a huge market and claims a slice, which any founder can type into a cell. Bottom-up builds revenue from the number of leads, conversion rates, price, and churn. It is slower to build and much harder to argue with, which is exactly why investors trust it more.
What is the most common mistake in a fundraising model?
Setting the raise or the valuation first and then bending the model to justify it. Investors reverse-engineer assumptions for a living, so a model built backward from a desired answer reads as exactly that. Build the drivers from the ground up, let the raise fall out of the runway math, and the whole document becomes more persuasive.