Free guide
Fundraising kit, guide 5 of 6Your financial model is the plan in numbers.
Investors do not check your forecast, they check your thinking. Nine sheets, every number traced to an assumption, and a runway you can defend in the meeting.
- Bottom-up, never top-down
- Every number traces to an input
- Monthly for 24 months
- Runway in months
At a glance
Nine sheets, one story
In the order you build them: the inputs, the customers, the revenue, the costs, the people, the spend, the cash, the what-ifs, the one-page summary. Tap a sheet to jump to it.
Sheet by sheet
What each sheet must do
The question it answers, what goes in, what stays out, and a test to check it passes.
Every input in one place: prices, conversion rates, churn, salaries, start dates, the raise. Each with a source or a reason. Every other sheet reads from here and hard-codes nothing.
Build in
- Every driver, one row each, in its own colour
- The source or the reasoning beside each number
- Start month, raise amount and opening cash
- A date on the sheet, and a version number
Leave out
- Numbers typed into formulas elsewhere
- Assumptions copied from a template you cannot explain
- Precision you do not have: 3.7% when you mean about 4%
Pass test: Change one input here and every sheet moves; nothing else needs touching.
The acquisition build: how many people you reach each month, how many convert, how many stay. Channel by channel, with the cost of each, so acquisition cost and payback fall out of the sheet.
Build in
- Each channel: reach, conversion, new customers per month
- Spend per channel and cost per customer
- Churn or retention, so the customer count is a stock, not a running total
- Payback period, from the cost above and the revenue below
Leave out
- Growth as a percentage per month with nothing behind it
- One blended acquisition cost with no channel
- Customers that never leave
Pass test: You can say where next month's customers come from and what each one costs.
Model rule: You sell subscriptions. Build customers as cohorts with monthly churn, add net revenue retention, and report MRR, payback and gross margin. Annual plans arrive as cash upfront and revenue over twelve months.
Customers times price, product by product, with the timing of cash: monthly, annual upfront, or on delivery. Built up from the customer sheet, never typed in.
Build in
- Customers by plan or product, from the sheet above
- Price per plan, from assumptions
- Timing: when cash arrives versus when revenue is earned
- Discounts, free tiers and trials as their own lines
Leave out
- A revenue line typed straight into the summary
- Top-down: a share of a market size
- Price rises you have not tested
Pass test: Revenue in any month equals customers times price in that month, and you can point at both.
Model rule: You have no revenue yet and few real data points. Spend the effort on drivers, hiring and runway, not on the revenue curve. Show the first customers as named pilots, and mark every price and conversion rate as a hypothesis.
The costs that grow with every customer: hosting, payment fees, support, delivery, third-party licences. Gross margin comes out of this sheet, and investors compare it to your category.
Build in
- Every cost that scales with customers or revenue
- Per-unit costs from assumptions, times volume
- Gross margin by month and by product
- Payment and platform fees as percentages
Leave out
- Salaries in cost of sales, unless the people deliver the product
- A margin copied from a public company
- Forgetting that the free tier costs money
Pass test: Gross margin lands in the range your category expects, or you can explain why not.
Model rule: You make a physical product. Add inventory and cost of goods with their timing: you pay suppliers months before customers pay you. Model the deposit, the production run and the cash gap explicitly.
One row per role: title, start month, fully loaded cost. People are the largest cost of most early companies, so the hiring plan is where the model gets honest.
Build in
- Every role, with its start month
- Fully loaded cost: salary plus tax, benefits and equipment
- Founder salaries, stated plainly
- The trigger for each hire: a milestone or a customer count
Leave out
- A headcount number with no roles behind it
- Hiring everyone in month one
- Salaries below what the market pays
Pass test: You can say who starts in month nine, and why then.
The costs that do not scale with customers: tools, rent, legal, accounting, insurance, travel, marketing programmes. Small lines that add up to a real number.
Build in
- Software and tools, one line each above a threshold
- Legal, accounting, insurance and compliance
- Office or coworking, travel, events
- Marketing spend, if it is not already in the customer sheet
Leave out
- A single miscellaneous line
- Costs that appear only after the raise closes
- Forgetting the tax bill
Pass test: Every line is a real supplier or a real category you could invoice.
Opening cash, plus the raise, plus receipts, minus everything above. The month the number crosses zero is your zero-cash date, and runway is the count of months until then. This is the sheet investors open first.
Build in
- Opening cash and the raise, by month received
- Monthly burn and cumulative cash
- Runway in months and the zero-cash date
- The milestone reached before cash runs out
Leave out
- Revenue counted before it is collected
- A raise assumed on day one
- A model that never runs out of money
Pass test: You can name the month you run out of cash without a raise, and the milestone you hit before it.
Example
Opening cash £40K. Raise £600K in month 1. Burn £35K a month rising to £55K. Zero-cash month 19. Milestone before it: £30K MRR in month 15.
Three versions of the same model: base, slower, faster. Change only two or three inputs, and show what happens to runway. Investors want to see you know which assumptions matter.
Build in
- Base, downside and upside as switches on the assumptions sheet
- The two or three inputs each scenario changes
- Runway and zero-cash date under each
- What you would cut, and when, in the downside
Leave out
- Ten scenarios
- A downside that is still a hockey stick
- Changing every input at once
Pass test: You can name the one assumption that, if wrong, shortens runway the most.
The sheet investors read: revenue, gross margin, operating costs, burn and cash by month, with the three or four metrics your category lives by. Built from the sheets above, never typed. Put it first in the file.
Build in
- Monthly revenue, gross margin, costs, burn and cash, from the sheets above
- The metrics that matter for your model: MRR, customers, payback, gross margin
- Charts of revenue and cash, one each
- Assumptions version and date
Leave out
- Annual totals that hide the monthly shape
- Metrics you cannot recompute from the model
- Colour, arrows and commentary
Pass test: An investor reads this sheet alone and can ask a precise question about any number.
Make it yours
Pre-revenue, SaaS, marketplace, hardware
The default model fits a software company before or just after first revenue. Four cases change the rules, and a few sheets are optional.
Pre-revenue
When: You have no revenue yet and few real data points.
Change: Spend the effort on drivers, hiring and runway, not on the revenue curve. Show the first customers as named pilots, and mark every price and conversion rate as a hypothesis.
SaaS
When: You sell subscriptions.
Change: Build customers as cohorts with monthly churn, add net revenue retention, and report MRR, payback and gross margin. Annual plans arrive as cash upfront and revenue over twelve months.
Marketplace
When: You match two sides and take a cut.
Change: Model both sides: supply and demand, each with its own acquisition sheet. Revenue is transaction volume times take rate; report volume and take rate separately, never a blended line.
Hardware
When: You make a physical product.
Change: Add inventory and cost of goods with their timing: you pay suppliers months before customers pay you. Model the deposit, the production run and the cash gap explicitly.
Optional sheets
- Cap table. You have taken money before, or the round has a SAFE or a convertible in it.
- Use of funds. An investor asks where the raise goes: the same numbers as the hiring and spend sheets, grouped.
- Cohort table. You have six months of customers and can show retention by month of joining.
- Sensitivity table. One input dominates and a grid of its values against runway says it faster than three scenarios.
- Fundraising plan. This round is one of several: show the next raise, its trigger, and the milestones before it.
Build it with AI
Hand this framework to your AI
Building the model with ChatGPT, Claude or any AI? Point it at this guide so it builds bottom-up sheets instead of a hockey stick. The whole method is published as two files, built from the same data as this page.
Or paste this prompt into your AI and add your prices, costs and plans:
Act as a startup finance coach and follow the MrAdib financial model framework at https://mradib.com/financial-model/llms.txt. Ask me for my prices, channels, conversion and churn, team plan, costs, opening cash and the raise, then build my model sheet by sheet: the rows, the formulas in words, and the assumptions each one reads. Bottom-up only, every number traced to an input, runway in months, no invented figures.
Sources
Where this comes from
The sheets distil what the investors and operators who read the most early-stage models say they check. Read them.
Questions founders ask
How far ahead should the model go?
Monthly for 24 months, then yearly to year three or five if the round needs it. Nobody believes month 47. The monthly view for the runway of this round is what gets checked.
Do I need a model before revenue?
Yes, a small one. It shows how you think about cost, hiring and runway, which is what pre-seed investors are buying. Two sheets done honestly beat nine done for show.
Excel, Google Sheets, or a modelling tool?
Sheets or Excel. Investors want to click into cells and see formulas. Send a copy they can edit, not a PDF, and keep a version you never share.
How precise does it need to be?
Precise about the logic, rough about the future. Every number should trace to an input you can defend; the inputs themselves are estimates. Round to the nearest thousand and say so.
The fundraising kit
Six guides, one set of facts
The whole kit, in the order you build it. Every piece draws on the same facts, so none of them contradicts another.
Numbers you can defend.
A model an investor can click into wins more trust than a curve that goes up. Questions about your model? Say hello.