Cash & capital

Cash runway: why cash divided by burn is the wrong formula

By Assaf Schwartz · 10 September 2026 · 8 min read

Cash divided by burn is the first number a founder learns and the last one they should trust. It answers a question nobody is asking — how long would we last if the business froze exactly as it is today — and the business never freezes. Run the same inputs through sixty months of arithmetic and the answer moves, sometimes by years, and often in the direction you were not expecting.

The formula everyone uses

The back-of-envelope calculation is runway = cash ÷ monthly net burn. Take the balance, take last month's outflow, divide. It is fast, it is checkable in your head, and it embeds one enormous assumption: that next month's burn equals this month's.

The simulator's projection makes no such assumption. It runs sixty monthly periods and reports the last month in which cash is still above zero. Everything that changes burn is allowed to change it — recurring revenue compounds at the growth rate and decays at the churn rate, payroll drifts up with salary inflation, cloud and licence spend drift up with cost inflation, variable infrastructure tracks revenue, and any operating profit is partly routed back into acquisition.

On the shipped defaults the two methods do not merely differ in detail. Month one burns $18,036 against $750,000 of Opening cash, so the envelope answers 41.6 months. The projection answers that the business never runs out at all: cash bottoms at $526,346 in month 23, one month before operating profit crosses zero in month 24, and recovers to $1,310,256 by month 60. The envelope was not slightly pessimistic. It was answering a different question.

Why burn never holds still

Five forces act on the monthly number at once, and they do not point the same way.

• Net revenue movement. Organic MRR growth adds recurring revenue; monthly revenue churn removes it, and the difference compounds. At the defaults that is 4.0% against 2.8%, so gross profit is a growing number pushing burn down every single month.
• Salary inflation. Payroll is indexed at the annual Salary inflation rate, so the largest line in most cost bases rises whether or not you hire anybody.
• Cost inflation. Baseline cloud spend and licence seats are indexed separately. This matters most to organisations carrying a large fixed infrastructure floor.
• Variable infrastructure. The Cloud cost per $1k MRR control means growth arrives with its own cost attached, so revenue never reduces burn by the full gross profit it adds.
• Re-investment. Once operating profit is positive, the Re-investment rate control diverts a share of it into paid acquisition. That buys revenue, and it holds net profit — and therefore cash — below where it would otherwise sit.

In a growing business the first force usually dominates, and it points downwards. That is why the naive formula errs pessimistically more often than optimistically: it charges the company for a burn rate it will never repeat. In a shrinking business every force points the other way at once, which is why the same formula fails so much more dangerously there.

Where the two answers part

The size of the gap is not a constant. It depends entirely on whether revenue is outrunning the cost base, and the sign of the error flips with it.

Three distinct patterns sit in that table. The defaults and the Series A scale-up both survive: the envelope hands them a finite death date they never reach, because break-even arrives first. The churn-crisis preset goes the other way — the envelope promises 12.9 months, and the projection has cash negative by month 13, because at 8.5% monthly churn revenue is shrinking while payroll is not.

The instructive row is the fourth. Take the defaults and move only Monthly revenue churn to 5%. Month-one burn barely reacts — $19,950 against $18,036 — so the envelope still reports 37.6 months, essentially unchanged. The projection reports 22 months and a trough of -$2.4M. One input that is invisible in this month's bank statement has taken a solvent five-year plan and given it a cash-out date inside two years.

What actually moves runway

Load the seed-stage preset and vary one control at a time. Two things stand out: how little the runway figure moves, and how much everything behind it does.

Doubling churn from 3% to 6% changes projected runway by 2 months. Over the same range it deepens the trough from -$329.2K to -$1.5M and moves operating break-even from month 36 to never. Manage this business on its runway figure and you would conclude that retention barely matters. It matters more than anything else on the page.

The cost-side levers

Engineering headcount is the blunt instrument: one engineer fewer is worth 8 additional months, one more costs 4. Salary inflation looks like a macro assumption nobody controls, yet moving it from 4.5% to 10% deepens the trough by $335.1K and pushes break-even from month 48 to month 54. If your plan promises annual rises, that is a cash commitment rather than a compensation policy, and it belongs in the model.

Now look at the last column across the whole table. Every one of those changes — a person on or off the team, a wage assumption, the entire contractor budget, eight points of gross margin — moves the break-even month within a span of 12 months. Churn moved it to never. Cost discipline buys time, reliably and immediately, and that is genuinely worth having. It does not, on its own, produce a business that stops needing it.

Minimum cash, not runway

Runway is a date. Minimum cash is an amount, and the amount is what you have to raise.

The seed-stage preset makes the point cleanly. Its projected runway is 17 months, which sounds like a fundraising problem for next year. But operating profit does not turn positive until month 48 — 31 months after the money is gone — and the low point of the cash curve is -$744,379 in month 47. That deficit, not the runway figure, is the real financing requirement.

The month the trough falls matters as much as its depth. A business whose low point is month 23 of a five-year plan has a bounded problem with a visible far side. A business whose low point is month 60 — the last month of the horizon — is telling you the model has not found a bottom at all. What you are looking at is a slope, not a curve, and the honest reading is that the plan does not close.

Planning around the real number

Four rules follow, in descending order of how much argument they save later.

• Quote three numbers, never one. Projected runway, minimum cash, and the month it occurs. A board that hears only the first will ask the wrong follow-up question.
• Raise to clear break-even, not to clear the runway. The distance between the month cash runs out and the month operating profit turns positive is the size of the problem. On the seed-stage preset that distance is 31 months, and no amount of cash discipline closes it on its own.
• Stress the retention control first. Churn moves the trough far more than it moves the date, and the trough is what you have to fund. Growth and headcount come next; inflation assumptions matter more than their size suggests.
• Re-run the projection when the plan changes, not when the quarter closes. A hire, a price change and a churn miss all move the projection immediately and the bank balance many months later. The lag is exactly the danger.

None of this makes cash divided by burn useless. It is a decent sanity check on the current month, and when it and the projection disagree by a factor of two, an assumption somewhere is doing something you have not noticed. Treat the disagreement itself as the signal, then go and find which of the five forces is responsible.

The two milestones behind this arithmetic get their own treatment in operating break-even vs. cumulative payback, and the control that deliberately trades cash for growth once you are past break-even is covered in the re-investment rate.

Educational content, not financial advice. Figures here are illustrative and exclude taxes, financing and one-off items. Model your own numbers in the simulator and check them with a qualified accountant before acting.