How to Build a Startup Budget Template, Column by Column

A startup budget template needs four things in this order: the opening cash, revenue in layers by how certain it is, every commitment priced by what it would cost to stop, and the remaining spending released against evidence. The order matters because costs tend to rise faster than they fall, and much of the difference is signed into the budget.

Mark Anderson, Rajiv Banker and Surya Janakiraman measured that difference. Across 7,629 firms over 20 years, selling, general and administrative costs rose 0.55% for each 1% increase in sales and fell only 0.35% for each 1% decrease (Anderson, Banker & Janakiraman, 2003). Sales that rise 10% and then fall back leave those costs about 2% higher than where they started (our arithmetic). For a startup facing a bad year, the sort that matters is how long each cost takes to stop, and what stopping costs; calling a cost fixed or variable answers neither.

Exhibit 1. Five steps to a startup budget, in the order that protects the cash

Step	The question it settles	The usual mistake
1. Open with cash	How much can leave, and when	Starting from last year’s profit and loss
2. Layer the revenue	Which revenue can carry a commitment	One blended growth rate
3. Price the commitments	What stopping would cost this month	Calling rent and salaries fixed without an end date
4. Release the rest against evidence	What must be true before money moves	Spending a 12-month plan by the calendar
5. Set the floor and review monthly	Whether the company can still stop in an orderly way	A floor of three months of expenses

Source: CX Cash method, built on the research and filings cited below.

A startup budget template runs 12 monthly columns over seven blocks of lines

The five steps fill one sheet, its rows grouped in seven blocks, each filled from a different record (Exhibit 2).

Exhibit 2. The lines of a startup budget template, block by block

Block	Lines	Filled from
Opening cash	The bank balance on day 1	Bank statements
Revenue, in four layers	Contracted, renewing, pipeline and unidentified, each in the month its cash should arrive	Contracts, renewal and close rates, the target
Payroll	One line per person or planned hire: start month, salary, employer taxes, benefits	Offer letters, the hiring plan
Committed costs	Office lease, annual software, cloud minimums, insurance	Signed contracts
Flexible costs	Contractors, paid acquisition, travel and events	Plans, each with its release trigger
One-off costs	Equipment, deposits, legal fees	Quotes
Closing cash and floor	Net cash flow, closing cash, unwind total, headroom above it	Formulas

Source: CX Cash method.

Across the top run 12 monthly columns from today’s bank balance, a year total and four that carry the method. Actual results and the variance against budget are filled in at each monthly close. The other three record what binds each line (the term or notice left), the cash it would take to stop (the unwind column) and, for flexible lines, the result that releases the next tranche. The free annual budget template below the article holds revenue, cost-of-revenue and operating-expense lines with full-year budget, year-to-date actual and variance columns; the monthly and unwind columns go beside them, in a spreadsheet or in budgeting software.

Payroll gets its own block because it is most of the spending. Kruze Consulting, an accounting firm, found in more than $900 million of venture-backed startups’ spending that payroll-related costs, contractors included, came to 76% of operating expenses. In its breakdown across more than 100 funded startups, contractors took 12.5% of operating costs, employer payroll taxes came to 7.6% of wages and benefits to 8% (Kruze, 2024). These are one firm’s figures, not a random sample. Kruze puts a hire’s total cost 17% to 30% above salary and suggests budgeting employer taxes at 10% of payroll early in the year, before wages reach the caps. A template that lists salaries alone understates its largest block, and the spread is why each hire is better priced on its own line than with one multiplier.

How to build a startup budget when costs fall more slowly than they rise

Cost accounting long assumed that costs fall as readily as they rise. “A fundamental assumption in cost accounting is that the relation between costs and volume is symmetric for volume increases and decreases”, Anderson and his co-authors wrote, before showing that overheads break the rule. Rajiv Banker, Dmitri Byzalov and Lei Chen then tested what makes the adjustment expensive: across firms in 19 OECD countries from 1990 to 2008, the degree of stickiness varied with the strictness of each country’s employment protection law (Banker, Byzalov & Chen, 2013). Both studies cover listed companies, and neither measures startups. The mechanism they point to, contracts and notice periods that make cutting expensive, binds any company that signs them. A startup also has less room for the asymmetry, because a cost that will not come down is paid out of a fixed stock of cash.

German accounting has named the effect since the 1920s. Kostenremanenz describes costs that, when activity declines, do not fall as far as they rose when it increased. German accounts date the term to Hans Brasch in 1927 and list notice periods and long supplier contracts among its legal causes (our translation; Kostenremanenz, German Wikipedia). The word comes from physics, where remanence is the magnetisation left in iron after the external field is removed (Remanence, Wikipedia).

Remanence is one symptom of a wider property: “Hysteresis is the dependence of the state of a system on its history”, so the same input gives a different output depending on the direction it came from (Hysteresis, Wikipedia). The shape carries over to a company’s costs; the cause does not. A magnet’s loop is a property of its iron. A company’s loop is a stack of signatures (leases, annual contracts, notice periods, hires), and most of them are signed in the budget or because of it. A budget therefore fixes two numbers at once: what the company spends if the plan works, and what it still owes if the plan fails.

WeWork sold flexibility and bought 15-year commitments

The second number has a public record at full scale. WeWork’s 2019 registration statement described a business built on “moving fixed lease costs to variable costs for our members”, who could take space “by the minute, by the month or by the year” (The We Company, S-1, 2019). The company kept the fixed side for itself. The average initial term of its US leases was about 15 years, and its future minimum lease payments under signed leases came to $47.2 billion at 30 June 2019. The average membership commitment, by contrast, had just risen to more than 15 months.

Measured against what members had bound themselves to pay, the gap was wider still. Committed revenue backlog, the non-cancellable part, stood at $4.0 billion, so the leases committed WeWork to almost 12 times what its members had committed to it (our arithmetic). In 2018 the company lost $1.93 billion on revenue of $1.82 billion. The S-1 named the risk in its own words: if departing members were not replaced, “our lease cost expense may exceed our revenue”.

On 6 November 2023, WeWork filed for Chapter 11 and asked the court for “the ability to reject the leases of certain locations” (WeWork press release, 2023). The Bankruptcy Code lets a debtor, with court approval, reject an unexpired lease (11 U.S.C. § 365). It then caps the landlord’s claim for future rent at the greater of 1 year or 15% of the remaining term, up to 3 years (11 U.S.C. § 502(b)(6)). On a lease with 12 years to run, that is 1.8 years of rent, or 15% of what was owed. Shareholders came last: under the restructuring support agreement, existing equity interests were to be cancelled with no distribution (WeWork 8-K, 2023).

A seed-stage company can repeat that structure in miniature: a commitment whose term runs longer than the revenue that pays for it. The budget is where the two terms sit side by side before either is signed.

Revenue enters the budget in layers, and only the firmest layer pays for commitments

Founders’ revenue forecasts tend to lean high, and in one large study writing them down made them lean further. Gavin Cassar found that nascent entrepreneurs whose ventures reached operation “overestimate the expected future sales and employment”, and that “the preparation of projected financial statements results in more overly optimistic venture sale forecasts” (Cassar, 2010). A budget is a projected financial statement, so the bias comes with the exercise. Cassar reads his evidence as causal; the safer reading is that the people who plan in detail are also the most optimistic, which leads to the same precaution.

The precaution is to enter revenue in four layers, each with the month its cash should arrive, instead of as one growth rate:

1. Contracted: signed, with billing dates and payment terms.
2. Renewing: existing customers due to renew, at the company’s own renewal rate.
3. Pipeline: named opportunities, at the company’s own close rate and sales-cycle length.
4. Unidentified: whatever else the target requires.

The matching rule follows from WeWork in reverse. Commitments (a lease, an annual contract, a multi-year cloud commitment) are funded only from the first two layers, and for no longer than those layers run. The third and fourth layers pay for spending that can stop within a month or two. If the target is met, nothing is lost, because the flexible spending runs all year. If it is missed, the shortfall lands on lines that can be stopped. The fourth layer, the gap between a top-down target and the bottom-up build, is the most exposed to the optimism Cassar measured, which is why it should never carry a lease.

Every commitment gets an exit price in the unwind column

The third step adds one column to the budget. We call it the unwind column: for every line, the notice period or term left, and the cash that would still go out if the company decided today to stop. Fixed and variable describe how a cost moves with sales; the unwind column describes how it moves with a decision.

For payroll, the exit price depends on where people work. The US federal WARN Act bars a plant closing or mass layoff until 60 days after written notice, but its definition of an employer starts at 100 employees (29 U.S.C. § 2101; § 2102). For a 12-person US startup, the cost of a layoff is set mainly by its own severance policy, before any state rules or contract terms. In the UK and Germany, a statutory minimum notice applies to individual dismissals and grows with service.

Exhibit 3. Minimum notice an employer must give, by jurisdiction

Jurisdiction and rule	Applies to	Minimum notice from the employer
US, WARN Act (29 U.S.C. §§ 2101–2102)	Plant closings and mass layoffs, at employers with 100 or more employees	60 days
UK, Employment Rights Act 1996, s.86	Employees with 1 month or more of continuous service	1 week under 2 years; 1 week per year of service from 2 to 12 years; 12 weeks from 12 years
Germany, BGB § 622	Employees	2 weeks during a probation of up to 6 months; then 4 weeks to the 15th or the end of a month; 1 month to month-end after 2 years, 3 months after 8 years, up to 7 months after 20 years

Sources: 29 U.S.C. §§ 2101–2102; Employment Rights Act 1996, s.86; BGB § 622 (our translation). Statutory minimums only: contracts, collective agreements, state law and German dismissal-protection law can add more.

In a small US company, payroll need not be the largest entry in the column. Consider two hypothetical US software companies with 12 employees and the same team. Company A has an office lease with 26 months to run, software on annual terms and a cloud contract with a committed minimum. Company B rents coworking space month to month, pays for software monthly and buys cloud capacity on demand, at slightly higher prices.

Exhibit 4. The unwind column for two hypothetical companies, $ thousands

Line	A: monthly cost	A: what binds it	A: cash to stop	B: monthly cost	B: cash to stop
Payroll, 12 people	195	4 weeks’ severance (company policy)	180	195	180
Office	22	Lease, 26 months left, no break clause	572	27	27
Software	9	Annual terms, 9 months left	81	11	11
Cloud hosting	18	$12k a month committed minimum, 8 months left	96	21	0
Contractors	24	30 days’ notice	24	24	24
Paid acquisition	30	Stops when the campaigns stop	0	30	0
Travel, events, other	12	One sponsorship already booked	15	12	15
Total	310		968	320	257

Source: our calculation for two hypothetical companies. Severance: 4 weeks of a $195k monthly payroll. Cash to stop is undiscounted and assumes no sublet or negotiated release.

The office lease is 7% of monthly spending and 59% of the cost of stopping. Payroll, 63% of A’s monthly spending, is 19% of the column. Company B pays $10k a month more, $120k a year, and in return its cost of stopping is $711k lower. Neither number appears in a startup budget template that lists only monthly costs, and the second is the one a downturn sends the bill for.

The cash floor is the cost of stopping, not three months of expenses

Brex’s and inDinero’s guides on how to build a startup budget tell founders to hold about three months of expenses in cash (Brex; inDinero). That rule asks almost the same of both companies: $930k for A and $960k for B. The unwind column asks A for $968k and B for $257k, a ratio of almost 4 to 1. Most of the gap is the lease. A better floor is the cash that lets the company stop in an orderly way, paying notice and honouring contracts, without a new round. The fifth step keeps the budget’s lowest monthly cash balance at or above the unwind total, and tests every new commitment against it before signing, which is where pre-approval for commitments belongs.

The test bites hardest on a single signature. Suppose company A starts the year with $5.0M, collects $80k in the first month and grows collections 4% a month; on those assumptions its cash stays above its cost of stopping until month 30. Now suppose that in month 6 it signs a five-year lease on a larger office at $40k a month and cannot sublet the old one. Monthly spending rises by 13%; the unwind total rises by $2.4M on the day of signing, and cash falls below the cost of stopping in month 10. A runway figure registers only the extra $40k a month; the date until which the company can stop in an orderly way moves 20 months closer.

The same test sorts the flexible spending of the fourth step. Lines that stop within a month (paid acquisition, contractors, events) can be budgeted for the year and released by the quarter, each tranche tied to a stated result: a payback period, a pipeline figure, a signed contract. The release rule needs a gap between its on and off thresholds. Engineers add hysteresis to thermostats on purpose, because it “prevents unwanted frequent switching” (Hysteresis, Wikipedia). A rule that releases paid acquisition while payback is under 12 months, and pulls it only once payback exceeds 18, will not switch a programme off in a bad month and on again in the next.

Then the review. At each monthly close the unwind column is updated, because lines change category without a meeting: an annual contract that renews automatically moves from flexible to committed on its renewal date, which makes every renewal date a decision date.

Anderson and Banker read sticky costs as a deliberate bargain

The researchers who measured stickiness did not call it a failure. Anderson and his co-authors model it as managers’ deliberate choice, and Banker, Byzalov and Chen read their evidence as showing “deliberate resource commitment decisions of managers in the presence of adjustment costs”. Cutting a team in a dip and rehiring in the recovery can cost more than carrying it through. Long commitments can also buy lower prices; WeWork’s own model rested on them, with an internal target of a 30% contribution margin “over the course of a 15-year lease” (S-1, 2019). In Exhibit 4, company B pays $120k a year for its flexibility.

Their reading holds on three counts. The cheapest budget is rarely the most reversible one. A company with multi-year contracted revenue can carry commitments of the same length. And in the US, payroll is cheap to unwind legally and expensive to unwind in practice, which is the managerial stickiness Anderson and his co-authors describe.

None of the three tells against the unwind column, which does not forbid commitments; it prices them and caps their total by the cash. A commitment that the cash can unwind is a decision. One that it cannot is a bet on the next round, and the budget should label it as one.

Putting the unwind column to work

1. Fill the startup budget template in the order of Exhibit 1: cash, revenue layers, commitments, flexible spending, floor.
2. Add an unwind column to whatever holds the budget, and update it at every monthly close. The board should see the unwind total next to runway, with its three largest lines.
3. Fund commitments only from contracted and renewing revenue, and for no longer than that revenue runs.
4. Keep the lowest monthly cash balance at or above the unwind total, and test each lease, annual contract and committed-spend agreement against it before signing. Take those above a set size to the board as separate decisions, with their effect on the floor shown.
5. Know how much of monthly spending could be stopped within 90 days. In Exhibit 4 the answer is 86% for A and all of it for B.

A spending plan and a list of what the company still owes

A startup budget template holds two documents in one: a spending plan if things go to plan, and a list of what the company will still owe if they do not. A budget that lists only monthly costs records the first. Companies fail on the second, because costs come down more slowly than they go up, and the difference is written into leases, contracts and notice periods before the year starts. For company A, one five-year lease signed in month 6 added $2.4M to that list on the day of signing and brought its last orderly stopping date 20 months closer.

Sources and notes

• Anderson, M. C., Banker, R. D. and Janakiraman, S. N. (2003). Are selling, general, and administrative costs “sticky”? Journal of Accounting Research 41(1), 47–63.
• Banker, R. D., Byzalov, D. and Chen, L. (2013). Employment protection legislation, adjustment costs and cross-country differences in cost behavior. Journal of Accounting and Economics 55(1), 111–127.
• Brex. How to create a healthy startup budget in 6 steps.
• Bürgerliches Gesetzbuch (BGB), § 622 Kündigungsfristen bei Arbeitsverhältnissen.
• Cassar, G. (2010). Are individuals entering self-employment overly optimistic? An empirical test of plans and projections on nascent entrepreneur expectations. Strategic Management Journal 31(8), 822–840.
• Employment Rights Act 1996 (UK), section 86.
• inDinero. How First-Year Startups Can Build a Useful Business Budget.
• Kruze Consulting (2024). Startup Payroll Costs and Tax Budgeting for New Hires.
• The We Company (2019). Form S-1 Registration Statement, filed with the SEC on 14 August 2019.
• United States Code, title 11, §§ 365 and 502(b)(6) (Bankruptcy Code); title 29, §§ 2101–2102 (Worker Adjustment and Retraining Notification Act).
• WeWork Inc. (2023). Form 8-K and Exhibit 99.1, WeWork Takes Strategic Action to Significantly Strengthen Balance Sheet and Further Streamline Real Estate Footprint, 6 November 2023.
• Wikipedia. Hysteresis; Remanence; Kostenremanenz (German).

The findings of Anderson et al. (2003), Banker et al. (2013) and Cassar (2010) come from their published abstracts, the WeWork figures from its 2019 S-1 and 2023 filings, the payroll shares from Kruze Consulting’s own data, and the notice periods from the statutes listed. Figures marked “our arithmetic”, and any other figure without a reference, are worked out from the referenced numbers, and the companies in Exhibit 4 are hypothetical. Translations from German are ours and carry no quotation marks.