Every business I have worked with has numbers. Ledgers, returns filed on time, a trial balance, a clean audited set of accounts. What most of them do not have is a model.
The distinction is not a technical one. An owner holding only the accounts is, quite precisely, driving by the mirror — and the mirror is showing a stretch of road finished nine months ago.
The Accounts
Describe what already happened. Governed by law, prepared for outsiders, accurate to the rupee, and available once the year is over. A scorecard.
The Model
Describes what would happen if. Governed by nothing, prepared for you, useful rather than exact, and available before you commit. A steering wheel.
What a model actually is
Strip away the software and a financial model is four things. A set of assumptions about the future. Arithmetic that links them together. Statements that emerge at the far end — a profit and loss account, a cash flow, a balance sheet. And a decision the whole apparatus exists to serve.
Only the first block is opinion. Everything after it is arithmetic that anyone can check. That is the most useful property a model has, and the one most often wasted, because it moves an argument to where it belongs. Two people disagreeing about a projected profit are never really disagreeing about the profit. They are disagreeing about an assumption three rows up that nobody has said out loud.
A model does not predict the future. It tells you what has to be true for your plan to work — and that is a far more useful object, because you can go and find out whether it is true.
1. Start with three drivers, not thirty
Most businesses rest on two or three numbers. Volume, price, and the rate at which one converts into the other. Everything else — rent, salaries, wastage, freight — either follows those numbers or sits still while they move.
The temptation, once a spreadsheet is open, is to add rows. Detail feels like rigour. It is usually the opposite: a model nobody can explain in two minutes is a model nobody will trust, and that includes the person who built it. Complexity hides the assumptions that matter behind twenty that do not.
Find the handful of numbers everything depends on. Then defend them. If you cannot say where a figure came from, you do not have a plan — you have a wish with decimal places.
2. Separate what you assume from what you calculate
In a well-built model the assumptions are one colour and the formulas are another. It sounds like housekeeping. It is closer to a discipline of honesty.
The moment the two are mixed, a hard-coded number buried inside a formula becomes invisible, and an argument that should have been about evidence becomes an argument about who sounds more confident. Keeping them apart forces every judgement into the open, where it can be questioned, sourced, and if necessary corrected without dismantling anything else.
It also makes the model survivable. Assumptions age; arithmetic does not. If the two are tangled, next year you will rebuild from scratch rather than update.
3. Ask what has to be true, not what will happen
This is the habit that changes how you read your own plan.
“We will make ₹40 lakh next year” is a forecast. You cannot check it, argue with it, or act on it — you can only wait and find out. Turn it around and it becomes something else entirely: at this margin and this cost base, we need this many units, at this price, from this many customers. Each of those is a claim about the world, and every one of them can be verified by somebody this month.
That inversion is most of what management accounting is for. It converts a number you hope for into a list of things you can go and confirm — and it usually reveals that one item on the list is doing all the work, and nobody has checked it.
4. Model cash before you model profit
Profit is an opinion — it depends on depreciation policy, on provisions, on when revenue is recognised. Cash in the bank depends on nobody's judgement.
The two diverge for a reason that has nothing to do with accounting mischief: timing. Money leaves when you buy material and pay wages. It returns when a customer settles, which may be sixty or ninety days after that. In a seasonal or growing business the gap between those two events is where the working capital goes — and it is entirely possible to be profitable on every rupee of sales while running out of money to make the next one.
Profitable companies close every year. Cashless ones close faster. A monthly cash projection is worth more than an annual profit forecast, and it is the document I would rather see first.
5. Test the answer before you believe it
Take the number your model depends on most, move it twenty per cent, and watch what happens — not to the profit, but to the decision. If the decision holds, you have room. If it flips, you have found the assumption that actually matters, and you now know exactly what to go and verify before committing money.
This exercise also exposes cost structure. A business carrying heavy fixed costs amplifies every movement in sales, upward and downward alike. Once that amplification is visible you can design around it — renting on a revenue share rather than a flat monthly figure, staffing peaks rather than the whole year. Converting fixed cost into variable cost lowers the point at which you break even and buys margin for error. That is a decision a model produces and an annual account never will.
The reason any of this matters
A balance sheet describes one day. For most businesses that day is 31 March, chosen for reasons of law and tax rather than truth — and for a seasonal business it is very often the calmest day of the year. Stock has run down, customers have paid, the overdraft has been cleared. The accounts are accurate and the picture is flattering, and both of those things are true at once.
The difficult month is somewhere in the middle of the year, and no statutory statement will ever show it to you.
That is the whole case for building a model: not to produce better numbers, but to see the months nobody reports on, early enough to do something about them.
