Financial Modeling: Building a Model That Survives Questions

That's why a model is judged on the clarity of its assumptions rather than the beauty of its numbers. A model whose assumptions can't be seen reads as a wish. One whose assumptions are visible and debatable builds trust even where the investor disagrees with some of them.

1. First rule: separate assumptions from calculations

Dedicate one sheet to every changeable assumption, and never hard-code a number inside a formula. Every calculation cell references a defined assumption.

The benefit is twofold: anyone can change one assumption and see its effect immediately, and the conversation with an investor becomes about whether the assumption is right rather than whether the arithmetic is right — a far more useful conversation.

2. The revenue engine: build it bottom-up

The most familiar error is building from the top: "the market is worth X, and we'll take 1%." That isn't a model; it's a wish.

Sound construction begins with actual activity:

Visits or leads × conversion rate = new customers
New customers + retained customers from the prior month = total customers
Total customers × average revenue per customer = revenue

Every figure in that chain is a debatable assumption tied to an activity you can influence. That's what makes the model a management tool rather than a marketing document.

Don't omit churn. A model assuming every customer stays forever produces impossible numbers within two years.

3. Costs: fixed and variable

Variable costs rise with each additional customer: cost to serve, payment fees, usage-linked hosting, support.
Fixed costs aren't tied to short-term volume: salaries, rent, subscriptions, statutory obligations.

Salaries are typically the largest line, and should be built from a hiring plan rather than a lump figure: who is hired, when, and at what fully loaded cost — not base salary alone.

Contribution margin = revenue − variable costs. This is what actually remains to cover fixed costs, and it's the measure that reveals whether the model is viable.

4. Cash flow: the decisive statement

Profit isn't cash. A company can show a profit and stop operating because it didn't collect on time.

What the model must reflect:
- Collection timing: when does the customer actually pay after invoicing?
- Payment timing for suppliers, salaries, and statutory obligations.
- Advance payments where they exist — a substantial cash advantage in annual subscription models.
- Capital expenditure where applicable.

The two most important outputs of the entire model: monthly burn rate and runway — the months remaining at current cash.

5. Scenarios

Three are enough: conservative, base, optimistic. The difference between them isn't "double everything" but changing two or three variables only — conversion rate, acquisition speed, and churn, for instance.

The question the conservative scenario must answer: do we survive if it happens? If the answer is no, that's more important information than anything in the optimistic case.

6. Horizon and granularity

• First 18–24 months: monthly, in full detail. This is the period actually managed.
• Beyond that to three years: quarterly or annual, at lower granularity.

Five-year projections at early stage are an exercise in imagination more than a tool. Include them if asked; don't base a decision on them.

7. Credibility test

Before presenting the model, test it against these questions:
- Can I justify every assumption with a source, an experiment, or a logic?
- Is assumed growth consistent with the hiring and spending plan?
- Did I calculate acquisition cost inclusive of marketing and sales salaries?
- Does churn appear?
- Does cash stay positive every month in the base scenario?
- What happens if revenue is delayed six months?

Common mistakes

• Building top-down from an assumed market share.
• Perfect linear growth with no setbacks and no churn.
• Hard-coded numbers inside formulas that can't be traced.
• Ignoring collection timing and its effect on cash.
• Estimated salaries with no hiring plan.
• Omitting statutory, accounting, and legal costs.
• A conservative scenario that is actually optimistic.

Checklist

• A separate, referenced assumptions sheet
• A bottom-up revenue engine
• Churn built into the model
• Fixed costs separated from variable
• A hiring plan feeding the salary line
• A cash flow statement with realistic timing
• Burn rate and runway visible
• Three scenarios with defined variables
• Monthly granularity for the first 18–24 months

FAQ

How many years should the model cover?
Three is usually enough, with monthly detail for the first two.

Should I use a template?
As a starting point, yes — but the revenue engine must reflect your business model, not the template's.

What most destroys a model's credibility?
Unjustified sharp growth, absent churn, and an acquisition cost that excludes salaries.