How to build a financial model for investors (and what they actually check)
A founder once sent me a 12-tab Excel file. Revenue projections, hiring plan, cohort charts, a sensitivity grid, the works. Three minutes into the call, the investor asked one question: "What's your burn multiple at the end of year two?" Silence. The file had everything except the number that mattered.
So here's the thing I keep coming back to: a financial model is not a spreadsheet. It is the shortest version of your business that an investor can argue with. If they can't argue with it, they don't trust it. If they can't find the assumptions, they assume the worst. Building a model that survives due diligence is less about Excel tricks and more about deciding what you're willing to defend.
This is the framework I use when I help founders prepare for a raise. It won't teach you spreadsheet shortcuts. It will tell you what a VC opens first, what makes them close the file, and how to structure your model so the numbers do the talking.
Key takeaways
- A financial model for investors exists to answer one question: can this business return the fund?
- Three statements, one driver sheet, no orphan numbers. If a figure has no formula behind it, delete it.
- Investors reverse-engineer your growth rate from your own revenue build. Round numbers and hockey sticks get flagged in seconds.
- Working capital and burn multiple kill more deals than weak revenue forecasts.
- Scenarios matter less than the honesty of your base case. One believable case beats three fantasy ones.
What investors actually open first in your financial model
It's almost never the profit and loss statement. Experienced analysts go straight to the cash flow tab. Then the assumptions sheet. Then they trace one revenue line backward to see whether it's hardcoded or calculated.
Why the cash flow tab comes first
Profit is an opinion. Cash is a fact. An investor who has watched a company die with a healthy P&L will check your runway before anything else. They want to see the month the cash balance goes negative without a new round, and the month it goes negative with one. If those two dates don't appear anywhere in your model, you're not finished.
I watched a seed-stage founder present a model where runway was a single cell, typed as "18 months." No formula. When asked what happened if hiring slowed by 30%, he had no answer. The round fell apart two weeks later, not because the business was bad but because the model couldn't be interrogated.
The assumption sheet is where trust is won or lost
Put every number you invented on one page. Every conversion rate, every churn figure, every salary band. Color-code the cells that are inputs versus calculations. It takes an afternoon. It changes how the whole conversation goes, because the investor can now test your logic instead of guessing at it.
Look at this contrast:
| Model element | What gets rejected | What earns the follow-up call |
|---|---|---|
| Growth rate | "30% monthly" with no driver | Built from leads × conversion × deal size |
| Headcount plan | One line: "team: 40 by year-end" | Role, start month, fully-loaded cost |
| Churn | A flat 2% forever | Cohort table showing improvement by vintage |
| Working capital | Missing entirely | DSO and DPO separate from revenue |
| Scenarios | Three tabs, identical assumptions | One case, clearly labeled drivers for upside |
The backbone: three statements that must reconcile
Every credible model I've reviewed has the same skeleton. Income statement, balance sheet, cash flow statement. Linked. In that order, feeding each other.
Build the income statement from drivers, not from percentages
New modelers reach for growth percentages. "Revenue grows 8% per month." That's a conclusion, not a mechanism. A driver-based build asks: how many leads, what conversion, what average contract value, how many sales reps, what quota per rep. Then revenue falls out of the math.
- Leads per month, split by channel
- Conversion rate per channel (and it will not be the same across channels)
- Average contract value, and how it shifts as you move upmarket
- Sales capacity: reps × quota × ramp curve
The ramp curve matters more than founders expect. A rep hired in January doesn't produce at full quota until month four or five. If your model has every new hire hitting plan immediately, an investor notices within ninety seconds.
Link the balance sheet or nothing holds
Skipping the balance sheet is the single most common shortcut I see. It feels tedious. It isn't optional. Without it, working capital has nowhere to live, and working capital is where a lot of attractive-looking businesses quietly run out of money.
Suppose you sell annual contracts billed upfront. Your cash looks fantastic. Your balance sheet carries deferred revenue, a liability you'll service for twelve months. An investor who sees the cash spike without the corresponding liability knows the model is decorative.
Cash flow is the reconciliation, not an afterthought
If net income plus changes in working capital doesn't equal the movement in your cash balance, you have a bug. Find it before the investor does. I'll admit that on my first startup model, my cash flow statement was off by roughly $40,000 and I spent two evenings hunting a sign error in a prepaid expense line. That kind of afternoon is part of the job.
How to build a financial model for a startup: revenue first, everything else second
Startups don't have history, so the model is a set of beliefs. The revenue section is where those beliefs are visible.
Bottom-up beats top-down, and it isn't close
Top-down math sounds like this: "the market is $4 billion, we capture 1%." Investors discount it to zero. It tells them nothing about your operations. Bottom-up math sounds like this: "we close 12 deals a month at $18k ACV by Q3, because we'll have five reps at full quota." That's testable. It's also falsifiable, which is why it earns attention.
Model churn by cohort, not with one blended number
Blended churn hides your best and worst customers in the same figure. Cohort tables break that. Group customers by the month they signed, then track retention per group. If your older cohorts retain better, you have a story about product maturity. If they retain worse, you have a story about a market you're misreading. Either way, you know. Investors trust the second kind of founder.
Worked example: a $2.1M ARR plan in one page
Take a subscription business targeting $2.1M ARR by month 24. Assume 40 new customers a month by month 18, ACV of $4,500, and monthly logo churn of 1.5% declining to 1.1% as onboarding improves. That gets you close to the target, and every input is defensible. Now stress it: drop new customers to 28 a month and hold churn flat. What's the ARR at month 24? If the answer is "still $1.4M," the plan is resilient. If it's "$600k," you've just learned which lever the whole story depends on.
That single test tells an investor more than three scenario tabs ever will.
Mistakes that kill the raise
The recurring failures aren't exotic. They're the same six or seven problems, over and over.
- Hockey stick with no driver. Growth accelerating without a headcount or channel change behind it.
- Round numbers everywhere. Every assumption landing on 10%, 20%, 30% signals the model was written backward from a target.
- Working capital ignored. Particularly in inventory or enterprise sales businesses.
- Expenses as a percentage. "Marketing at 20% of revenue" hides whether the spend can actually acquire the customers you forecast.
- One model, no version control. When the investor asks what changed since the last call, you should know.
The burn multiple deserves its own line. It's new cash burned divided by net new ARR in the same period. If you spend $1 to add $1 of ARR, that's a 1.0. Investors have different thresholds by stage, but the direction is universal: below 1 is efficient, above 2 starts a conversation you don't want to have.
Scenarios, sensitivity, and the trap of over-modeling
One honest base case beats three fantasy cases
I used to build aggressive, base, and conservative tabs. Then I noticed investors only ever read the aggressive one, assumed it was the base, and discounted everything else accordingly. Now I present one case with clearly labeled drivers, plus a short note on what would push it up or down. Less work, better conversations.
When sensitivity analysis is worth the effort
Sensitivity grids earn their place when one variable dominates the outcome. If a 10% change in churn swings your runway by six months, show that. If nothing moves the needle much, a grid is decoration. Use judgment. Not every model needs a data table.
Common questions about investor-ready models
How to build a financial model for investors in Excel?
Excel remains the standard, and for good reason: investors want to manipulate your assumptions themselves. Structure it with one tab per statement, one dedicated assumptions sheet, and no formulas that reach across more than two tabs. Use consistent formatting, freeze your panes, and never hardcode a value inside a calculation. Google Sheets works for early conversations, but expect to export to Excel before a serious diligence process begins.
Is there a free way to build one?
Yes, and you don't need paid software. Start from a template, strip it down to your business, and rebuild the revenue section yourself so you understand every formula. The rebuild is the education. Templates that you don't understand are worse than a blank sheet, because they create false confidence.
The model is a conversation, not a verdict
Here's what I've come to believe after watching dozens of these processes: investors don't fund accurate forecasts. Nobody can forecast a startup accurately. They fund founders who clearly understand which assumptions their business depends on, and who know what breaks first when those assumptions move.
So when you open the spreadsheet tonight, don't start with the revenue line. Start with the question you're least sure about. Model that one first. The rest gets easier, and the conversation gets a lot more interesting.
