A model an investor can actually check.
Financial modelling for a raise. A driver-based model built from your own numbers, where every assumption can be traced back to something real when an analyst starts pulling at it.
Why most fundraising models get opened once
It usually goes like this. A founder downloads a template a week before the first investor meetings, types in 15% monthly growth, fills the cost lines from last month's P&L, and the thing looks finished. Three tabs, a chart, a runway number. Then an associate asks where the 15% comes from, and the honest answer is that it felt reasonable.
Nothing in that model is wrong arithmetically. It just can't be explained, and a model that can't be explained is worse than no model, because it tells the investor something about how the founder makes decisions.
Here's the part founders don't always hear. Investors rarely believe your forecast. Nobody at a fund expects you to hit your year three revenue. What they use the model for is working out how you think the business runs: which levers move revenue, which costs grow with it, how much cash the plan really needs, and whether the amount you're raising is connected to any of that or picked first.
So a fundraising model is less a prediction than an argument. It should be roughly right and completely explainable. The work is making it that.
What gets built and handed over
- A driver sheet
- Every input in one place, each labelled with where it came from: your actuals, a signed contract, a benchmark with a named source, or a judgement call marked plainly as one. This is the sheet investors spend the most time on.
- A revenue build from the real unit
- Built from whatever actually drives revenue in your business. Customers, seats and price for software. Billable hours, utilisation and realisation for services. Orders and average order value by location for food service. Patient volume and revenue per episode for healthcare.
- A cost build that knows what scales
- Headcount by role and month at fully loaded cost, including employer PF, gratuity accrual and ESIC where it applies. Other costs split between the ones that grow with revenue and the ones that don't, because that split is where margin arguments are won or lost.
- Three statements that tie
- P&L, balance sheet and cash flow, linked so that cash is a result rather than a typed number. GST and TDS timing are carried where they move cash enough to matter.
- Use of funds and runway
- What the raise pays for, month by month, and when the money runs out in each scenario. Written so the ask in your deck and the model say the same thing.
- Scenarios you can defend
- A base case, a case where your most important driver underperforms, and what you would cut if it did. Not a best case. Investors build their own.
- A metrics sheet
- The numbers funds ask for at your stage, calculated from the model rather than typed beside it: gross and contribution margin, CAC payback, burn multiple, and ARR where it applies.
- An assumptions note
- A short document explaining each material assumption in plain words. It's what you send when the analyst's questions arrive, and it saves a week of calls.
Seed and Series A models are different jobs
| Seed | Series A | |
|---|---|---|
| What investors test | Whether you understand what drives the business | Whether those drivers held up against actuals |
| History behind it | Often under a year, sometimes none | Usually 18 to 24 months of monthly actuals |
| Revenue build | Bottom-up from early signals: pilots, pipeline, conversion on a small base | Cohorts, retention and acquisition cost by channel |
| Level of detail | Fewer drivers, each argued well | More drivers, each reconciled to history |
| Where it usually fails | Growth nobody can source | Assumptions that ignore what the last 18 months showed |
At seed, thin history is expected and forgiven. What isn't forgiven is pretending otherwise. A 36-month curve built on two months of revenue says more about the founder's judgement than about the business, and seed investors are mostly buying judgement.
At Series A the question turns round. There is now enough history to test the plan against, and the fund will test it. If the model assumes 4% monthly churn and the last six months averaged close to 7%, someone will find it. That conversation goes far better when you raised the gap yourself and explained what changed.
One assumption, traced through a raise
An example of why the driver matters more than the growth rate. The company and every figure here are illustrative.
A B2B software company has ₹38 lakh of monthly recurring revenue and wants 20 months of runway from its next round. The founder's template grows new customers 12% a month and keeps the sales team at four people.
Rebuilt from drivers, new customers come from salespeople. Over the last two quarters a fully ramped rep closed about three deals a month, and a new hire took around four months to get there. Growing new customers 12% a month therefore means hiring reps well ahead of the revenue they'll bring, and paying them in the meantime.
| Template version | Driver-based version | |
|---|---|---|
| Sales reps by month 12 | 4 | 9 |
| Monthly burn in month 12 | ₹41 lakh | ₹58 lakh |
| Cash needed for 20 months | ₹7.6 crore | ₹10.4 crore |
| Can the founder explain the growth? | No | Yes, rep by rep |
The template didn't raise too little because its maths was wrong. It raised too little because growth was typed in instead of paid for. A founder who closes ₹7.6 crore on that model discovers somewhere around month 14 that they need to raise again, and they do it with less runway and less room to negotiate than they had the first time.
The driver-based version isn't more cautious. It's connected. When an investor asks what happens if ramp takes six months instead of four, the answer is one cell away, and you can show it in the meeting instead of promising to come back.
Where fundraising models go wrong in India
Most modelling advice online is written for American companies. A few things are specific here, and each one moves the cash number.
- Cash is treated like profit. GST collected on your invoices isn't yours to spend, input credit arrives later, and TDS your customers deduct sits with the tax department until you claim it back. For a services business with large clients, ignoring this can understate the cash a plan needs by more than a month of payroll.
- Salary is used as the cost of an employee. Employer PF, gratuity accrual and ESIC where it applies all sit on top, and under the labour codes in force since November 2025 the wage base they're calculated on may be higher than it was in your old salary structure.
- The model runs on calendar years. Indian companies report on an April to March financial year, and investors will set your model beside your audited accounts. Calendar-year models turn a simple check into a reconciliation exercise.
- Dollars and rupees are mixed. Software companies selling abroad often quote ARR in dollars and pay costs in rupees. Without one reporting currency and a stated exchange rate assumption, margins move with the rupee and nobody reading the model can tell why.
- The raise amount comes first and the plan is built to spend it. Investors notice when runway lands on exactly 18 months in every scenario.
None of these is exotic. They're just easy to leave out of a template that was built somewhere else.
Where the model sits in the rest of the raise
The model is not a separate deliverable that lives in its own folder. Every number in your deck should come out of it: the revenue on the traction slide, the margin you quote, the runway the raise buys. When a deck says 62% gross margin and the model says 57% because one was calculated before hosting costs and the other after, an investor doesn't see a rounding difference. They see two versions of the business.
The same goes for the data room. The historical months in the model should match the monthly MIS you share, which should match the books, which should match the bank. That chain is what a diligence team walks, usually backwards from the bank. Building the model is often the first time anyone walks it from the front.
So the work tends to surface things. A revenue line recognised on invoice rather than delivery. A contractor cost sitting in the wrong month. Better found now, by you.
What we'll ask you for
- Monthly P&L for as far back as it exists, from the books if possible rather than rebuilt from memory.
- Bank statements or a cash summary, so the P&L can be reconciled against what actually moved.
- Current headcount with roles, joining dates and CTC.
- Revenue split the way your business really earns it: by customer, product, plan or location.
- The cap table and the terms of any earlier rounds.
- Your own view of the plan, in plain words. What would you do with the money? The model's job is to test that view, not to replace it with ours.
If the books are a few months behind, that's normal at seed and it's fixable. We'll tell you plainly what the model can be built on and what needs cleaning first, because a model sitting on numbers that don't reconcile fails in diligence no matter how good it looks.
Questions people ask first
Related on this site
Sources
Checked in September 2026. Rules, rates and published figures change, so confirm anything you act on with your CA, lawyer or payroll provider.