Budget and Runway Model

01 · People

Seventy to eighty percent of the spend, and the section where the arithmetic errors are largest.

• Contractors costed separately, with end dates

  Each contractor with a rate, a scope, and the month the engagement ends.

  Usually missing: The end date. Contractors get budgeted as though they run for the whole project, or they get budgeted for three months and quietly extend. Both produce a number nobody can plan against.

  Done when

  • Every contractor line has a start and an end month.
  • Anyone who has already extended past their original end date is shown at their real duration, not the planned one.
  • The total contractor spend is visible as its own number.
• Cost by role and month, not by department

  Every person and every open role, with a monthly cost and the month they start.

  Usually missing: The month. A budget that says "three engineers, 2.4M" hides whether they start in January or September, which is a difference of nearly a year of salary. Department totals are what a finance team reports; role-by-month is what a producer can actually plan against.

  Done when

  • Every row has a start month, and the months are different for at least some of them.
  • Someone who is not on the team could read the sheet and say how many people are being paid in any given month.
  • Roles you intend to hire are visibly separate from people you already have.
• Employer costs on top of salary

  Social fees, pension, insurance, and vacation accrual, as a stated percentage on top of gross salary.

  Usually missing: The multiplier entirely. In Sweden the loading is roughly 31.42 percent in employer contributions before pension and insurance, so a budget built on gross salaries alone understates the people cost by a third. It is the single most common arithmetic error in a small studio budget.

  Done when

  • A stated percentage is applied to every employed person, and the number is written down rather than embedded in a formula.
  • Contractors are costed at their invoice rate, without the employer loading applied.
  • Someone has checked the percentage against a current source rather than against memory.
• Founder compensation, in the budget

  What the founders draw, from which month, at what rate — including zero.

  Usually missing: The month it starts. A budget where founders are unpaid forever is not a plan, it is a hope, and an investor reads the missing line as a cost that will appear after the money arrives. Say when it starts and what it is.

  Done when

  • Every founder has a rate and a start month in the sheet.
  • Any deferred or accrued pay appears as a liability, not as a saving.
  • The budget still works when founder pay begins.

02 · Production

Engine, outsourcing, platform, localization. The games-specific costs no generic template asks about.

• What is treated as investment and what as cost

  A stated position on which development spend is capitalized, agreed with your accountant.

  Usually missing: A position at all. It changes the shape of the accounts an investor reads and how a grant treats your spend, and most small studios have never had the conversation. It is not a budgeting question but it lands in the same sheet.

  Done when

  • Your accountant has stated a position in writing.
  • The budget reflects it consistently rather than mixing both treatments.
  • Anyone reading the sheet can tell which treatment is being used.
• Engine, middleware, and tools, with their terms

  Every licensed thing in the project, with annual cost and any revenue share or royalty.

  Usually missing: The royalty. A five percent engine royalty above a threshold is not a cost line in the budget — it is a haircut on revenue that never appears in the spending plan and surprises everyone at launch. Model it in the revenue side or it will be modeled for you.

  Done when

  • Every engine, middleware, plugin, and paid tool appears with a cost.
  • Any royalty or revenue share is stated with its threshold and rate.
  • The list was built from the project files rather than from memory.
• Localization and age ratings

  Languages chosen with a reason, costed per word, plus rating board fees per territory.

  Usually missing: Ratings entirely. IARC covers most digital storefronts at no cost, but a physical release or certain territories require paid board submissions, and teams discover this in the final month. Localization gets budgeted for translation and not for the integration and testing that follows it.

  Done when

  • Each language has a per-word cost and a stated reason for being included.
  • Integration and linguistic QA are budgeted separately from translation.
  • Rating requirements are confirmed for every territory you intend to ship in.
• Outsourced work, with a real rate and a real volume

  Art, audio, VO, QA, and porting priced by volume at quoted rates, not as a lump sum.

  Usually missing: The volume. "Art outsourcing: 800K" is a wish. Three hundred props at a quoted per-asset rate is a budget, and the difference shows up as a forty percent overrun about two thirds of the way through production.

  Done when

  • Each outsourced category has a unit, a volume, and a rate.
  • At least one quote from an actual vendor backs the rate, rather than an industry average.
  • There is a stated assumption about how many revision rounds are included.
• Platform, certification, and store costs

  Devkits, developer accounts, certification submissions, and the store cut, each as a line.

  Usually missing: Failed certification. Everyone budgets one submission. Console certification commonly takes two, sometimes three, and each round is calendar time with the whole team waiting — which costs far more than the submission itself.

  Done when

  • Devkits and developer accounts appear per platform.
  • The budget assumes at least two certification submissions.
  • The store cut is in the revenue model, not omitted.

03 · Runway

When the money runs out, and what you have reached by then.

• The date an orderly wind-down is still possible

  A named date, before zero, that accounts for notice periods and final obligations.

  Usually missing: The whole concept. Founders plan to zero and discover that closing a company costs money — notice periods, final invoices, accrued vacation — so the real decision point is one to three months earlier than the balance suggests.

  Done when

  • A date is written down, and it is not the same as the zero date.
  • It accounts for notice periods for everyone employed.
  • The founders have agreed what decision gets made on that date.
• Burn as a curve, not an average

  Monthly outflow for every month of the plan, showing the peaks.

  Usually missing: The peak. An average burn of 400K hides the three months at 700K when outsourcing, certification, and a marketing beat land together — which is the month the company actually runs out of money, several months before the average said it would.

  Done when

  • Every month has its own number and they are not all the same.
  • The highest month is identified and someone has checked the balance holds through it.
  • Lumpy costs land in the month they are actually paid.
• When the raise starts, backward from the floor

  A start date derived from how long a raise actually takes, not from when the team feels ready.

  Usually missing: The duration. A seed round takes three to six months from first conversation to money in the bank, and teams start when they are two months from zero — which is a negotiating position, and not a good one.

  Done when

  • The raise start date is on the same timeline as the burn curve.
  • It assumes at least four months, and says so.
  • There is a stated fallback if the raise takes longer.
• Runway measured to a milestone, not to zero

  The date the money runs out, and the milestone reached before it does.

  Usually missing: The gap between them. Running out of money one month after a vertical slice is a fundable position; running out one month before it is not, and the difference is often a single hiring decision made a year earlier.

  Done when

  • Both dates are on the sheet — money-out and milestone-reached.
  • The gap between them is stated in weeks.
  • There is a written answer to what happens if the milestone slips a month.

04 · Risk

What breaks the plan, and whether you held anything back for it.

• Contingency, named and sized

  A stated percentage held back, with a rule for when it can be spent and who decides.

  Usually missing: The rule. A contingency with no spending rule is spent by March, usually on something reasonable, and then there is no contingency for the thing it existed for. Ten to fifteen percent is a normal figure; the discipline matters more than the number.

  Done when

  • A percentage is stated and visible as its own line.
  • There is a written rule for when it may be drawn and who approves it.
  • It has not already been allocated to planned work.
• Currency and rate exposure, if you have any

  Costs and revenue in their real currencies, with the rate assumed stated once.

  Usually missing: Any acknowledgment. A Swedish studio paying an outsourcer in dollars and earning in dollars while paying salaries in kronor has exposure in both directions, and a ten percent move is a real number on a multi-year plan. Nobody needs to hedge; everyone should know the sensitivity.

  Done when

  • Costs are recorded in the currency they are actually paid in.
  • The assumed rate appears once as a named figure, not in each formula.
  • Someone has looked at what a ten percent move does to the total.
• What breaks the plan, and by how much

  Two or three named variables with the effect of each on the runway.

  Usually missing: The ranking. Teams model three scenarios by scaling everything up or down together, which tests nothing. Changing one variable at a time tells you which one to watch, and it is usually a hiring date rather than a revenue assumption.

  Done when

  • At least three variables are moved independently and the runway effect is recorded for each.
  • The variables are ranked by impact.
  • The top one has a named owner watching it.

05 · Discipline

A budget nobody reopens is a document. This section is what makes it a tool.

• Actuals compared to plan, monthly

  Last month's real spend against the budget line, reviewed on a fixed day.

  Usually missing: The comparison. Budgets get built once and never opened again, so the first time anyone compares plan to reality is when the bank balance surprises them. A monthly hour is the whole discipline.

  Done when

  • The last three months each have actuals recorded next to the plan.
  • The review happens on a fixed day and someone owns it.
  • At least one budget line has been revised as a result.
• One person owns the model

  A named person who maintains the sheet and is asked before it changes.

  Usually missing: The name. A budget maintained by everyone is maintained by nobody, and the version in the investor deck stops matching the version the team plans against — usually without anyone noticing which one is wrong.

  Done when

  • One person is named, and the team knows who it is.
  • There is one file, and everyone can say where it lives.
  • The version in any deck was exported from it rather than retyped.
• A trigger that forces a reforecast

  A named condition — a variance threshold or a milestone slip — that requires the plan to be rebuilt.

  Usually missing: Any trigger. Plans get reforecast when someone gets nervous, which is later than it should be and correlated with bad news rather than with real change. A ten percent variance or a slipped milestone is a mechanical trigger and it removes the judgment call.

  Done when

  • A threshold is written down.
  • Someone is named as responsible for calling it.
  • It has been triggered at least once, or the plan has genuinely held.