What the model is for
Nobody opens a seed financial model (finansmodell) to learn what the company will earn in year five. It is opened to see how you reason: what you believe drives the business, how those beliefs connect, and whether the amount you are asking for follows from anything.
Three checks get run, usually within a few minutes. Run them yourself before the file goes out:
- Change one assumption and watch what moves. If nothing moves, there is no model, only a picture of one.
- Trace the ask back to the cash line. The round size should be the number that carries the company to a stated point.
- Compare the model against the deck and against the accounts. Three documents describing the same company should agree.
An investor’s scorecard covers the same ground, and an investor who cannot answer it from your model marks it down without saying why. Its economy section asks whether a budget exists, whether there is a milestone plan, and whether the funding need has been estimated. The funding-round section asks for:
- the pre-money valuation against the last round
- soft and hard commitments received
- whether existing shareholders are participating
- the use of proceeds
- the runway the round buys
- the size of the round after this one
Structure: drivers first, statements second
A model that survives questioning has one place where beliefs live and one place where arithmetic lives, on separate sheets. Keep the assumptions sheet the only one you ever type into.
| Sheet | What belongs there |
|---|---|
| Assumptions | Every input a human chose: prices, conversion rates, churn, salaries, hiring dates, payment terms. Nothing calculated. |
| Revenue build-up | Volume times price, built from the drivers that actually produce sales, not a growth percentage applied to last month |
| Costs and headcount | One row per role with a start date, the full employer cost per role, and non-payroll costs tied to the driver that causes them |
| Cash | Timing: when invoices are paid, when payroll leaves, when holiday pay and tax fall due, when grant money arrives |
| Statements | Profit and loss, balance sheet and cash flow as outputs of the sheets above |
The three statements are outputs. Typed in directly, the model has no mechanism, and the first question an investor asks will break it.
Model monthly for the first 18 to 24 months, then quarterly or annual. This is a rule of thumb with no external standard behind it. The first 24 months are the period the round actually funds, and due diligence reads current-month cash.
The assumptions investors will test
Five assumptions carry most of the weight, and each is judged on whether it is defended, not on whether it is optimistic. Prepare a one-line defence of each before the meeting.
- Growth, and its justification. Where does the next customer come from, through which channel, at what cost.
- Gross margin, and its defence. What sits in cost of sales, and why the margin holds as volume rises.
- Customer acquisition cost against customer lifetime value. Both need to be calculable from the model’s own drivers.
- Churn. Its level, and whether the revenue build-up actually applies it.
- Hiring pace. Headcount is the largest cash line in almost every early-stage Norwegian company, and the one founders most often understate.
The Norwegian cost lines a model must carry
An employee costs more than their salary. A model that expenses gross salary and stops understates cash by a wide margin and misstates its timing.
| Cost line | The rule, and what it does to the model |
|---|---|
| Employer’s national insurance contributions (arbeidsgiveravgift) | Rate by zone: 14.1 per cent in zone I, 10.6 per cent in zone II, 6.4 per cent in zone III, 5.1 per cent in zone IV, 7.9 per cent in zone IVa, 0 per cent in zone V. Zone Ia is 14.1 per cent above a tax-free amount (fribeløp) currently around NOK 850,000 per enterprise, within which 10.6 per cent applies. The zone depends on where the enterprise is registered and operates; most Oslo-area companies are zone I. It is levied on the whole payroll cost, not on base salary alone. |
| Holiday pay (feriepenger) | Statutory minimum 10.2 per cent of salary; 12 per cent where five weeks’ holiday is agreed, which is common in Norwegian employment contracts; 12.5 and 14.3 per cent respectively for employees over 60. Earned in one year and paid the next, normally in June, so it is a liability accrued in the year of work and a cash outflow in the year after. |
| Mandatory occupational pension (obligatorisk tjenestepensjon) | At least 2 per cent of the employee’s income from the first krone and up to 12 G (OTP-loven § 4), for employees who are compulsory members of Norwegian national insurance, are at least 13 years old and earn above a small reporting limit, currently around NOK 2,000, which Skatteetaten publishes. A standing monthly cost from the first employee. |
The safe formulation for a seed model is salary, plus holiday pay at 10.2 to 12 per cent, plus mandatory pension at 2 per cent, with employer’s national insurance contributions on top of the total.
Grants and SkatteFUNN
SkatteFUNN gives a tax credit of 19 per cent of approved research and development costs, for companies of any size. The cost basis is capped at around NOK 25 million per company per income year; Skatteetaten publishes the current cap.
Own employees are costed at 1.2 per mille of agreed annual salary per hour, capped at roughly NOK 700 per hour and about 1,850 hours per employee per year, with the current limits on the same Skatteetaten page. A company that is not in a tax position (the normal case for a loss-making startup) receives the credit as a cash payment through the tax settlement (skatteoppgjør).
The timing is what the model gets wrong. Forskningsrådet approves the project; Skatteetaten grants the money through the company’s tax return; the cash therefore arrives with the following year’s tax settlement. Applications submitted by 1 September are guaranteed processing within the year and give credit for that income year.
Show it as its own cash line, arriving late, and discount anything not yet approved. Grant income modelled as customer revenue, in the month the work is done, overstates cash and shortens your real runway by the same amount.
The red flags that end the reading
This list and the next have been stable for years. Build the model against them before you send it.
| Red flag | What the investor concludes |
|---|---|
| Hardcoded numbers where formulas should be | There is no model. The numbers were decided and then typed in |
| A hockey-stick curve with no driver behind it | The founder is describing an outcome they want rather than a mechanism they have |
| Costs that stay flat while revenue multiplies | Either the business is misunderstood or the cost side was never built |
| Payroll modelled as gross salary only | The founder has not run a Norwegian payroll and has not asked anyone who has |
| An ask unconnected to break-even or to the milestone plan | The round size came first. Everything after it is presentation |
| Model numbers that contradict the deck | One of the two is wrong, and both are now suspect |
| A single scenario with no stated sensitivity | Nothing has been tested, including by the founder |
What a healthy model shows
Each of these is a claim the investor can test by changing a cell. Make sure the cell exists, and that the change shows in the cash line.
| Signal | What it demonstrates |
|---|---|
| Growth that can be justified driver by driver | The founder knows where customers come from |
| Gross margins that can be defended as volume rises | The unit economics are understood, not assumed |
| Low customer acquisition cost against high lifetime value | The business can afford to grow |
| Low churn | Customers who stay are evidence of value delivered |
| Trends that improve over the modelled period, for a stated reason | Learning is being priced in, not hoped for |
| A funding need that reaches break-even or the next milestone | The ask is arithmetic, not ambition |
Granularity and horizon by stage
Precision the business cannot support reads as false confidence, and investors are practised at spotting it. A five-year revenue line to two decimal places, built pre-revenue, damages credibility.
Pre-revenue, the honest model is short and driver-heavy: cost and headcount modelled properly, revenue as a small number of explicit assumptions you are willing to defend, and a clear statement of what the round funds.
Post-revenue, the historical months become the model’s calibration. Actuals versus plan, month by month, is a stronger argument than any forecast, because it shows your estimates being tested against reality.
Build two or three scenarios, each driven by changing one or two assumptions in the assumptions sheet, with the change visible. Three columns labelled conservative, base and aggressive, with differences typed in by hand, test nothing.
The model in the raise
Financial due diligence asks for last month’s bank statement, the current cash position and the burn rate. These are current-month figures, and they will be compared against the model without warning.
Historical numbers must tie to what is filed, and current numbers to the bank statement and the ledger. Filed annual accounts sit in the Register of Company Accounts (Regnskapsregisteret), where anyone can retrieve up to fourteen years of them without asking. The tax return (skattemelding) is not public and will be requested in the data room (datarom) instead.
A historical tab that does not reconcile to the filed accounts is a diligence finding before anyone has looked at the forecast. The model also has to stay usable through the months a raise takes:
- Keep one version of the truth. One file, dated, with the version shared with each investor recorded. Sending an updated model without saying what changed is worse than sending nothing.
- Answer model questions from the model. If a question cannot be answered by changing an assumption and showing the result, say that the model does not cover it. Do not improvise a new number in the meeting.
- Update the actuals monthly through the raise. A model whose last actual month is four months old tells an investor the company is not run from its numbers.
What to prepare, by situation
The checks are the same for every founder. What you prepare first depends on where you are in the raise.
- First raise, pre-revenue. Build the cost side first and build it completely: every role with a start date and the full employer cost from the table above, because headcount is the line investors expect you to understate. Keep revenue to the few assumptions you can defend out loud, and let the ask fall out of the cash line as months of runway to a named milestone.
- Second raise. Open with actuals against the plan you showed last time, month by month, with the misses explained. Reconcile the historical tab to the filed accounts before anything else, since the investor can pull them from Regnskapsregisteret without asking you. Show the last round’s money in the cash line: what it bought and how long it lasted.
- With a lead already committed. The lead will forward the model to co-investors and run the three checks on their behalf, so the assumptions sheet has to read without you in the room. Agree with the lead which dated version is the diligence version, freeze it, and log every change after that in one place.
- Without a lead. Several investors will open the model in the same weeks, and each will compare it with the deck and the bank statement. One file, one dated version per investor, and answers from the model only, so that no two investors hear different numbers for the same question.