Decide what the plan is for before you open the spreadsheet
A financial plan written to raise money and a financial plan written to run a business are different documents. The first is an argument. The second is an instrument. Trying to make one document do both is the most common reason plans end up useless for either purpose — too optimistic to manage against, too detailed to persuade with.
Be explicit about the audience. A lender is underwriting downside: can this business service the debt if things go moderately wrong? An investor is underwriting upside: how large could this get, and what has to be true? You, running the business, need something else entirely — a monthly comparison between what you expected and what happened, close enough to act on.
If a plan cannot be wrong, it cannot be useful. A forecast with no falsifiable assumption is a brochure.
The practical consequence: build the operating model first, honestly, and derive the lender or investor version from it. The reverse never works, because a document designed to persuade acquires optimism you then have to unpick before you can use it.
Three statements, and why they must connect
A financial plan is not a revenue forecast. It is three linked statements, and the links are the entire point — they are what stops you from projecting profit you have no cash to fund.
If the three are not linked, the model will happily show a profitable business running out of money without ever flagging it. That is not a hypothetical failure mode — it is the standard one for anything growing quickly.
Ending cash on the cash flow statement must equal cash on the balance sheet, in every period. If it does not, something is wrong and every number downstream is unreliable. Build that check into the model on day one and leave it visible.
Assumptions belong in one place, and should be graded
Every number in a forecast is either an assumption or the arithmetic result of one. Assumptions belong on a single sheet, each entered once, each labelled. Numbers typed directly into a calculation are the reason models cannot be updated six months later.
More useful than listing assumptions is grading them by how much you actually know. Reviewers — and you, in six months — need to know which inputs are observed and which are hopeful.
Twelve months of actuals$1,840Observed
Consistent across three years42%Observed
Aged receivables report48Observed
Signed lease$8,400/moObserved
Nine months of data, seasonal31%Partial
Small sample, improving64 daysPartial
Based on one hire$340K/yrEstimate
Trade association estimate9%Estimate
The discipline this creates is immediate: if your entire growth case rests on two assumptions graded Estimate, that is worth knowing before you present it, and it tells you exactly which two numbers to go and validate.
Build revenue from drivers, never from a growth rate
“Revenue grows 20 percent” is not a forecast. It is a wish with a number attached. It cannot be tested, cannot be managed, and tells you nothing about what to do differently if it does not happen.
Build it from the things you can observe and influence. The exact drivers depend on the business, but the structure is always the same — a volume, a price, and a rate connecting them.
| Business | Volume driver | Rate | Value |
|---|---|---|---|
| Professional services | Billable hours available | Utilisation | Rate per hour |
| Retail | Foot traffic or sessions | Conversion | Average basket |
| Subscription | New sign-ups | Retention | Monthly value |
| Practice | Patient visits | Case acceptance | Revenue per visit |
| Manufacturing | Capacity in units | Utilisation | Price per unit |
Built this way, the forecast becomes a set of statements you can check against reality every month. If revenue misses, you know whether it was volume, conversion or price, and those have completely different responses.
Whatever drives your revenue, ask what physically constrains it. A practice has chairs and hours. A manufacturer has machine time. A services firm has people. If the forecast implies more output than the constraint allows, the plan needs a capital line or a hiring line — and both consume cash before they produce it.
Costs: separate what scales from what does not
The distinction that matters is not fixed against variable in the textbook sense. It is which costs move with volume, which move in steps, and which do not move at all. Getting this wrong is why forecasts look plausible in the base case and fall apart in every other one.
- Variable — materials, merchant fees, direct labour where hours genuinely flex. Model as a percentage of revenue or per unit.
- Step — a supervisor at fifteen staff, a second location, another machine. These jump at a threshold and then stay flat. They are where naive percentage-of-revenue models break.
- Fixed — rent, insurance, core salaries, software. They do not care what you sold this month.
Headcount deserves its own schedule: role, start month, salary, and a loading for payroll taxes, benefits and workers’ compensation. The loaded cost of an employee is commonly twenty-five to forty percent above salary, and a plan that forecasts base salary alone understates its largest expense line by roughly a third.
A forecast where every cost is a fixed percentage of revenue will always look profitable at scale. It is also always wrong, because nothing real behaves that way.
Cash is the point, and growth consumes it
Profitable businesses fail. They fail because profit is recognised when work is delivered and cash arrives when the customer pays, and the gap between those two events has to be funded by someone. Usually you.
The mechanism is the cash conversion cycle: how long from paying for something to collecting for it. Every day in that cycle is working capital you have to finance, and it scales directly with revenue. Faster growth means more cash tied up, not less.
| Component | Meaning | Days |
|---|---|---|
| Days inventory | Held before sale | 42 |
| Days sales outstanding | Invoice to collection | 48 |
| Days payable | Your terms with suppliers | (31) |
| Cash conversion cycle | Days of funding required | 59 |
Fifty-nine days on $2.4M of revenue is roughly $388,000 permanently tied up. Grow revenue by a third and you need about $130,000 more — before a single new hire or piece of equipment. That number belongs in the plan explicitly, because it is the amount you must fund from profit, a line of credit, or your own pocket.
02Related toolWorking Capital CalculatorMeasure your cash conversion cycle and what growth ties up.→03Related toolLine of Credit QualifierEstimate a borrowing base against receivables and inventory.→Three cases, and the one that matters
A single-case forecast is a prediction, and predictions are wrong. Three cases turn the plan into a range and force you to say what you would actually do at each end. Vary the two or three assumptions that genuinely drive the outcome — not every input, which produces noise rather than insight.
What you actually expect. Every assumption at its most likely value, not its most hopeful. This is the case you manage against.
Note what the upside case does to cash. Higher revenue, higher EBITDA, and less money in the bank — because growth funded the receivables and inventory before any of it was collected. A plan that shows the upside case with more cash than the base case has not linked its statements properly.
The downside case, and specifically whether debt service coverage stays above about 1.25 times in it. A plan showing 3.0x coverage in the base case and 0.9x in the downside will be underwritten on the 0.9x. Build the downside deliberately rather than leaving a reviewer to construct their own, less charitable version.
The mistakes that show up most
- The hockey stick with no mechanism — flat history, then sharp growth beginning the month after the plan was written, with nothing in the cost lines to explain what causes it.
- Margins that improve for no stated reason — gross margin drifting up two points a year is a claim about pricing power or purchasing scale. Say which, or hold it flat.
- No working capital line — growth modelled with no funding requirement attached. This is the single most common structural error.
- Annual periods — a yearly model hides seasonality and cannot show the month you run out of cash. Model monthly for at least the first two years.
- Hardcoded numbers — assumptions typed into formulas rather than referenced from one sheet. Makes the model unmaintainable and errors invisible.
- No comparison to actuals — a plan never reopened after it was written taught you nothing. The value is in the variance.
The last one is the one worth changing first. A modest plan, compared against actuals every month and revised when it is wrong, is more useful than a sophisticated one written once and filed. The point was never the forecast. It was learning how the business actually behaves.