A model built from drivers, not guesses.

Open a typical startup model and the revenue line grows by a percentage each month. Ask where the percentage comes from and the honest answer is that it looked achievable. Everything downstream inherits that guess: the hiring plan, the costs, the runway, the amount being raised.

A driver-based model works the other way round. Revenue comes out of things you can observe and argue about: how many customers you have, how many you add, how many leave, what they pay, how many hours your team can bill and at what realised rate. Growth is then a result rather than an input, and every number can be traced back to something real.

That's what this template is. It isn't a three-statement model and doesn't pretend to be. It's an operating model for a company with subscriptions, services, or both, across 24 months.

The model is only as good as eight or nine numbers. Each should come from your own history where it exists, and be marked as a judgement where it doesn't.

1. 01Customers at the start and new customers in month one: from your billing system, counting customers who actually pay.
2. 02Growth in new customers: from the last six months, not from what the plan needs. If sales capacity drives it, remember the headcount sheet has to add the people.
3. 03Price: average revenue per customer after discounts, not list price. Schedule a price rise only if you have decided to make one.
4. 04Churn: from your cohort data, averaged over several months. This single number moves the 24-month picture more than any other.
5. 05Variable cost per customer: infrastructure, usage, payment fees and support. Estimate it honestly rather than leaving it at zero.
6. 06Billable hours and realised rate, for services revenue: from timesheets and invoices, not from the rate card.
7. 07Payment days: how long customers really take, from receivables history.
8. 08The loading on top of CTC: small if your CTC already includes employer PF and gratuity, larger if it doesn't.

Where an assumption is a judgement, write the reasoning in the notes column. When an investor asks, and they will, the answer is on the same screen as the number.

Hiring at salary, not cost

Employer PF, gratuity accrual, insurance and equipment all sit on top. The headcount sheet applies a loading so the plan carries the real number, which is what the bank balance will feel.

Everyone starting on day one

Hiring takes longer than plans assume. Put realistic start months in, then see what it does to cash. A role starting in month eight rather than month five is worth several lakh of runway.

Churn left out

A model that adds customers and never loses them overstates month 24 badly. Even modest churn compounds: at 3% a month, you lose about a third of a cohort in a year.

Cash equals profit

The cash sheet moves receipts by your payment days, which is why a plan that looks fine on EBITDA can still dip. It's a simplification, and it catches the main effect.

Ambition entered as an assumption

If the plan needs 12% monthly growth and your history says 4%, the model will happily produce the number. It just won't survive the first question about it.

The template opens with a fictional B2B software company so every formula shows something. It starts with 210 customers at ₹18,000 a month, adds 16 new customers in month one growing 4% a month, loses 1.8% of customers each month, and also sells services at 320 billable hours a month at ₹2,400 an hour. It has 12 people at the start and hires five more over two years.

Read down the cash line and you can see the shape the model exists to show: cash falls for the first year while the company invests, bottoms out around ₹1.44 crore, then recovers as EBITDA turns positive. Change churn from 1.8% to 3% and that recovery moves out by quarters. That sensitivity, visible in about ten seconds, is the reason to build a model at all.

Models produce confident-looking numbers from weak assumptions, which is why a few minutes of checking is worth more than a day of formatting.

1. 01The history check. Does month one of the model roughly match last month's actuals? If the model's starting point is already 20% away from reality, everything after it is decoration.
2. 02The capacity check. If revenue triples, who delivers it? Compare the revenue line with the headcount line and ask whether that team could serve those customers. Plans that grow revenue faster than the people to deliver it are the most common kind of implausible.
3. 03The reverse check. Take the month-24 revenue and work out what it implies: how many customers, how many new ones a month, how many salespeople, what churn. If the implied numbers are ones you would not defend in a meeting, change the assumptions rather than the conclusion.

That third check is the one investors run instinctively. A founder who has already run it answers questions calmly, because they have seen the same arithmetic.

• ·A balance sheet. This is an operating model, so there is no closing position for receivables, payables, equity or reserves.
• ·Tax. Most early-stage companies are loss-making, and tax on profits deserves its own treatment with your CA.
• ·Depreciation and capital spending, which need a fixed asset schedule to do properly.
• ·GST and TDS timing, which move cash in the short term. The 13-week cash forecast template handles that horizon.
• ·Fundraising. Add a row for money raised in the month you expect it, and see the effect on the cash line.
• ·Multiple currencies. If you sell in dollars, convert at a stated rate and keep exchange movement out of revenue.

For a seed or Series A conversation, an operating model of this kind plus an assumptions note is usually what investors actually want. Full three-statement models become necessary later, or earlier if your business has significant inventory, debt or capital assets.

Most fundraising models are opened twice: once while raising and once when someone asks about the next round. That's a waste, because the model is the cheapest early warning system a company has.

The habit worth forming is a monthly comparison. Put actual revenue, customers, churn and costs beside what the model assumed for that month. You are not looking for precision. You are looking for a driver that has been consistently off in the same direction for three months, because that is the one quietly deciding your runway.

When a driver has clearly moved, update the assumption, save a new version, and keep the original. Having the plan and the current view side by side is what lets a founder tell a board exactly how far the year has drifted and why, rather than presenting a model that has quietly caught up with reality.