Financial model and numbers
The financial model assumptions investors test
Short answer
Investors test six assumptions: customer growth, retention, price, acquisition cost, gross margin and hiring pace. For each, have the number, where it comes from and what happens to cash if it is 30% worse. A model is trusted when its assumptions tie to observed data.
The six
| Assumption | The question | Evidence |
|---|---|---|
| Customer growth | What produces new customers? | Sales capacity, pipeline, channel tests |
| Retention | How long do they stay? | Cohort table |
| Price | What will they pay? | Contracts, price tests |
| Acquisition cost | What does one cost? | Spend ÷ customers by channel |
| Gross margin | What does delivery cost? | Supplier and hosting bills |
| Hiring pace | Who do you add, when? | Role list with salaries and start months |
Run your own sensitivity check
Take each assumption and make it 30% worse, one at a time. Record what happens to cash at month 18.
| Change | Cash at month 18 |
|---|---|
| Base case | |
| New customers 30% lower | |
| Churn 30% higher | |
| Acquisition cost 30% higher | |
| Hiring three months earlier |
The two rows that move cash most are the ones to prepare for. They are also where you should spend effort gathering better data.
How to present assumptions
- One sheet, every input labelled with its unit
- A "source" column: actual, test result, comparable, estimate
- History beside forecast, so the reader sees the join
Answers that work
Weak: "We assumed 10% monthly growth, which is conservative."
Strong: "New customers come from field reps. Each rep closes eight a month by month three; we have six reps now and add one a month. That gives 48 to 80 new customers a month over the year."
The second answer invites a useful conversation. The first invites disbelief.
Assumptions founders forget
- Ramp time for new hires
- Seasonality
- Payment delays from customers
- Price pressure as competitors respond
- Taxes and compliance costs
Test yours in Model Builder: change an input and watch the lowest cash point.
Quick answers
How many assumptions should a model have?
As few as explain the business. Twenty clear inputs are better than two hundred.
What if I have no data for an assumption?
Say so, give the basis you used, such as a comparable company or a small test, and show the range.
Do investors expect me to hit the forecast?
They expect you to understand it. The model shows how you think; nobody believes month 30 to the rupee.