Most Financial Models Are Built to Be Presented. Build Yours to Be Used.

A founder sends across a fourteen-tab model ahead of a lender meeting. The revenue build is detailed, the charts are clean, the terminal value is calculated three ways. Then someone asks one question: if debtor days slip from 60 to 85 for two quarters, when do you breach the DSCR covenant? The room goes quiet and the answer arrives by email the next day. That model was built to be presented. It was not built to be used. The gap between those two things decides whether you walk out of the meeting with a term sheet or with a list of follow-ups.

Start From the Decisions, Not the Revenue Line

Before you open a spreadsheet, write down the five or six questions the model exists to answer. In practice they are specific: how much capital do we need and in which month does the requirement peak; at what revenue does the business stop consuming cash; how much term debt can this cash flow actually service; what happens to margins if we commission a second facility; what does an acquirer get for the price being discussed.

Each of those questions dictates structure. A working capital question needs monthly granularity, not annual columns. A debt capacity question needs a full repayment schedule with moratorium, amortisation and interest on average balances. A valuation question needs unlevered free cash flow and a cost of capital you can defend. Build the model to answer the questions you will actually be asked, and it will be smaller and more useful than the template version.

The most common failure is a model that forecasts everything and answers nothing. It has a three-statement structure because the template had one, twelve revenue lines because the business has twelve products, and no view at all on the two or three variables that decide whether the plan works.

Drivers Are the Model. Everything Else Is Arithmetic.

Every number belongs to one of three categories: an input, a driver, or a calculation. Inputs sit in one place, clearly marked, entered once. Drivers are the ten to fifteen variables that genuinely move the business. Everything else must be a formula.

Take a business selling through a field team. Revenue is productive headcount times accounts closed per rep per month times average ticket size times retention. That decomposition is testable. Someone can argue that a rep closes four accounts a month rather than six, and that argument is a real business conversation with real consequences for hiring and cash. A revenue line growing at 30 percent a year is not testable. It is an assertion dressed as a number.

Here is the discipline test. Change one input cell and the entire model should respond correctly, including the cash flow and the debt schedule. Count the hardcoded numbers buried inside formulas. The target is zero. Every hardcode is a place where the model will silently stop telling you the truth the moment an assumption changes.

If It Does Not Reconcile to Cash, It Is a Forecast

The three statements must link, and the balance sheet must balance in every period without a plug. The plug is where credibility dies. A credit analyst or a diligence team will find it within ten minutes, and from that point every other number in the file is treated as unverified.

Working capital is where mid-market models are most often quietly wrong. Receivable days should be driven off revenue, payable days and inventory days off cost of goods sold. Driving all three off revenue is a modelling shortcut that understates the cash absorbed by growth, sometimes materially. Keep the first twenty-four months monthly, because a business that is profitable for the year can still run dry in March when a large receivable slips and a tax payment lands in the same fortnight.

The debt schedule deserves the same rigour: opening balance, drawdown, repayment, closing balance, interest on the average balance, with term loans, working capital lines and unsecured borrowings kept separate. Then compute DSCR and interest cover in every period rather than showing a single comfortable figure on the summary tab. Covenants are tested periodically, so your model should test them periodically too.

Scenarios People Actually Believe

A sensitivity table that flexes revenue growth by plus or minus two percent is decoration. Businesses do not fail because growth came in at 18 percent instead of 20. They fail because several things moved together in the same unhelpful direction.

Build three cases, each with a one-line operating story behind it. Base is the plan management is willing to be held to. Downside should mirror what a lender will stress: revenue 20 to 25 percent below plan, the collection cycle extended by 30 days, a price concession to hold volume, and the additional working capital those three things demand. Upside is reserved for what you would do with more capital, not for optimism about the same plan.

Then look at the outputs that matter. In the downside case, what is the lowest cash balance, in which month, and how much covenant headroom remains at that point? If the answer is uncomfortable, you have just learned what you need to raise, in what form and by when, on your own terms rather than a lender's. That is the entire commercial value of the exercise.

A Model Nobody Maintains Is a Model Nobody Trusts

Give it one owner. Load actuals monthly and run the variance at driver level, not line level. Knowing revenue missed by 12 percent tells you nothing you can act on. Knowing it missed because rep productivity ran at 3.2 accounts instead of 5, while ticket size held, tells you whether to fix hiring, training or pricing.

Keep version discipline: date-stamped files and a short assumptions log recording what changed and why. In a live transaction you will be asked why the July model shows different EBITDA from the September one. A precise answer builds credibility. A vague one invites a discount on price or a tightening of terms.

Finally, keep it small enough that a competent outsider can follow it in an hour. If tracing a single number takes three tabs of mapping, the model will not be opened in a decision meeting and it will not survive diligence.

Apply the test to whatever sits on your drive right now. Ask a hard question about your own business: what a 45-day payment delay from your largest customer does to your borrowing requirement in the third quarter. If you can answer with the model open in front of you in under two minutes, and the person across the table can follow the logic as you do it, the model is doing its job. If you cannot, the fix is almost never more detail. It is usually less: fewer tabs, fewer forecast lines, sharper drivers, and a cash flow statement you would be willing to defend line by line.

Disclaimer

This article is for general information only and does not constitute financial, investment, tax, legal, or professional advice. The views expressed are for general informational purposes and should not be relied upon as a substitute for advice specific to your business or circumstances. Please seek appropriate professional advice before making any business, financial, investment, or transaction-related decision.