A model is not a forecast. It is the instrument you make decisions with.

Three uses, one structure

The same model serves three purposes, and they pull in different directions. For fundraising it has to survive an investor rebuilding it line by line. For planning it has to turn a decision into a number within an hour. For a board it has to explain what happened against what was expected, without the reader needing a walkthrough.

Most companies have three different spreadsheets pretending to be one model, and the differences between them are discovered by someone else.

What a model is actually for

Not prediction. Testing: what happens to runway if hiring slips a quarter, if churn halves, if the price rises ten per cent. A model that cannot be interrogated is a document, not a tool.

Structure before numbers

• A driver map. The eight to twelve variables that actually determine the business, stated once and used everywhere.
• Bottom-up revenue. Customers, units, price, mix, churn: the way the business makes money, not a percentage attached to last year.
• Unit economics as a live engine. Acquisition cost, delivery cost, retention, all pulling from the same drivers, so a change in one propagates.
• Hiring and capacity. People as a constraint, not a cost line: who must exist before revenue can happen.
• Cash, monthly. Profit and cash separate, because a profitable month with a payment delay is still a cash problem.
• Scenario switch. Base, downside and the specific decisions that differ, not a multiplier applied across the whole sheet.

The test of a good model

Hand it to someone who did not build it, ask for the effect of a ten per cent price cut on runway, and see whether they find it in minutes. If they have to reconstruct your logic, the model belongs to you rather than to the company.

The familiar set

• Revenue as one growth rate, which makes every downstream number an opinion.
• Costs as fixed percentages of revenue, which hides the step changes that decide survival.
• No cash flow statement, so profitability is confused with liquidity.
• One scenario, presented as the plan, without the decisions attached to the alternatives.
• Forty tabs with no input sheet, where changing an assumption requires editing formulas.
• A model built to impress during the raise and abandoned eight weeks later.

The last one is the most common and the most expensive. A model that is not used monthly stops being true within a quarter, and then it is only evidence that the company once had a plan.

What investors actually check

Two things, quickly: whether the revenue build is bottom-up and whether the cash balance matches the story of the raise. Everything else is read after those two pass.

Deliverables

• Model with a single input sheet, driver map and a scenario switch that changes decisions rather than percentages.
• Monthly revenue build to unit level, with cohort retention where the business has repeat customers.
• P&L, cash flow and a runway view, with the breakeven point visible rather than implied.
• Hiring and capacity plan tied to the revenue it is expected to produce.
• One-page assumptions memo: what was assumed, what was left out, and what would change the answer most.
• Handover session, so the model can be maintained by the team that owns it.

What we need from you

Twelve to twenty-four months of actuals where they exist, current contracts and pricing, hire plan, and the honest list of what is not yet working.

Four stages

Driver mapping with the founders, before any spreadsheet is opened. This is where most of the value is created.

Build, with the model tested backwards against historical actuals: if it cannot reproduce the past, it will not describe the future.

Scenario design, tied to decisions the company can actually take.

Review and handover, including a short guide to changing assumptions without breaking the structure.

Who does what

You own the commercial facts and the decisions. We own the structure, the arithmetic and the documentation. Your accountant owns the statutory reporting, which is a different artefact with different rules.

Where it stops

It does not make weak unit economics work. If the acquisition cost exceeds what a customer returns, the model shows it sooner, which is the point.

It does not value the company: valuation and unit economics.

It does not write the fundraising document it feeds: investment business plan.

It does not replace management accounts, and it is not a substitute for financial control.

Related work

Where the same numbers must answer a public funding panel's budget requirements: government and innovation funding.

Where this work leads

One document answers one reader, but the question behind it rarely stays inside a single direction.

• Growth Strategy

The level above

Documents solve a defined file. Where the decision itself is open, the work moves up a level, into a direction, and eventually into growth strategy.