Default Alive Is a Question Every Founder Should Answer Monthly

Default alive means one thing. On today’s cost base and today’s revenue growth, with no new money, you reach profitability before the cash runs out. Default dead means you do not.

Answer it monthly. The cheapest correction available to you takes roughly three months to execute, and a quarterly review can burn that whole margin before it tells you anything.

Paul Graham named the question in October 2015 and reported that half the founders he spoke to could not answer it.¹ Ten years on the question is unchanged. Three things underneath it are not: the gap between rounds, the currency your costs are priced in, and the share of your capital that is debt.

Why these three models

This piece runs the Wire Model: score the features of the decision, route to a small ensemble of formal models, then force the ensemble to produce dated actions. The scores that mattered:

• Threshold and tipping dynamics (0.9). Default alive is a boundary condition. Small moves in growth or burn flip the whole regime, and crossing the line changes how everyone else behaves toward you.
• Regime-break risk (0.75). Debt financing, currency resets and a stretched round gap have altered the rules Graham’s version assumed.
• Strategic actors with asymmetric information (0.7). You know your burn. Your investor knows your claim. Those are different objects.
• Historical-analog density (0.65). Graduation rates, round gaps and bridge shares are measured, so a base rate exists.
• Cognitive distortion (0.8). The burn number is produced by the person with the most to lose from it.

That routes to three models: threshold (why the flip is sudden), Markov state-transition (where companies actually go, and how often to look), and signaling (why the answer is priced). They span three outcome types, complex, cycle-regime and equilibrium, so their errors point in different directions and partly cancel. The behavioral layer scored high enough for a fourth seat. It rides inside all three cards instead, because it changes how you read every lever below and adds none of its own.

The framework: the correction lag decides the interval

1. The boundary is a cliff, and it is reflexive

Granovetter’s threshold model describes outcomes where each actor’s choice depends on how many others have already chosen.² Small shifts in the distribution of thresholds produce enormous shifts in the outcome. Nothing moves, then everything moves at once.

Runway behaves the same way. Below a certain months-to-zero, your counterparties start acting on the number rather than on the business. Suppliers shorten terms. Senior hires stop signing. Investors move you from the monitor pile to the wait-for-the-next-round pile. Every one of those decisions consumes cash, which lowers months-to-zero, which triggers the next one.

Graham gave the trap its name. The fatal pinch is default dead, plus slow growth, plus not enough time to fix it.¹ Read that third clause as the operative one. The first two conditions are recoverable. The third is what kills.

Graham left one part implicit, and most founders never price it. The fix has a duration. In Kenya, terminating a contract on account of redundancy requires notifying the employee and the labour officer at least one month before the intended date, paying one month’s wages in lieu of notice, and paying severance of no less than fifteen days’ pay per completed year of service.³ Add the selection process, the handover, and the committed spend you cannot exit inside a quarter. Call the whole thing the correction lag. For most teams it is about three months, and for the first of those months the cut costs you money rather than saving it.

Threshold, the boundary lens

Assumes: counterparty behaviour changes discontinuously once runway crosses a visible line.

Fits because: threshold and tipping dynamics scored 0.9.

Breaks when: your cash position is genuinely private and no counterparty can observe it. Rare, and it stops being true the moment you open a raise.

Counteracts: the belief that runway declines smoothly and can be managed smoothly.

May reinforce: panic cutting, and treating every dip as the cliff.

2. Where companies actually go, and how often to look

Treat the company as sitting in a state each month and moving between states with some probability: default alive, default dead with time, default dead without time, raised, closed. Base rates tell you the transitions. They are not encouraging and they are not secret.

Across roughly 50,000 startups on Carta, the median time from seed to Series A now runs about 2.1 years, up from 1.5 years in 2019, and the 2019 first-quarter cohort took four years to reach a 49.1% graduation rate.⁴ The two-year conversion rate fell from 30.6% for the first-quarter 2018 cohort to about 15.5% for the 2023 cohort.^{5, 6}

Now do the arithmetic. Median seed raise on that dataset is around 4 million dollars.⁴ Median wait to the next priced round is about 25 months, before you add the three to six months the raise itself takes. Budget 18 months of runway against a 25-month median gap and you are default dead by construction, on the median path, on day one. The plan did not fail. The plan was arithmetic that never closed.

The sampling rule falls out of the same structure. If your correction lag is roughly three months and you observe your own state every three months, then in the worst case you find the problem a full quarter after it appeared, having spent the entire window in which the cheap fix was still available. Observe monthly and detection costs you one month of runway at most. Measure it quarterly and you convert a decision into a postmortem.

The interval is the argument. Graham said to start asking after eight or nine months and to err on the side of too early.¹ The cadence itself follows from your correction lag, which you can measure in an afternoon from your employment contracts, your lease and your supplier terms.

Markov, the base-rate lens

Assumes: transition probabilities between states are stable enough to plan against.

Fits because: historical-analog density scored 0.65 and the graduation data is measured rather than anecdotal.

Breaks when: the matrix itself changes. The 2021 transition probabilities did not survive 2022, and anyone who planned on them was wrong for reasons that had nothing to do with their company.

Counteracts: planning off your best month instead of off the population.

May reinforce: fatalism, and applying a median to a company that is genuinely in the tail.

3. The answer is priced, so producing it is a signal

Spence’s result is that a signal separates types only when it is costly, and costly in a way that hurts the weak sender more than the strong one.⁷ That is why investors stopped taking growth claims at face value and started pricing efficiency.

David Sacks gave the market its shorthand in April 2020: burn multiple, defined as net burn divided by net new annual recurring revenue, a way to judge whether burn is too high in any given month, quarter or year.⁸ Note the unit. He specified month first.

The separation runs like this. Claiming default alive costs nothing. Becoming default alive costs you the roadmap, the two hires you wanted and possibly a market you were holding open. The weak sender will not pay that. So the act separates, and the artifact that proves it is a series of monthly readings computed the same way, not a slide produced during a raise.

The market has already priced the alternative. On Carta, bridge rounds rose to roughly 16.6% of all capital raised in the second quarter of 2025, from 11.8% a year earlier and under 10% across 2021, and reached 22.5% of all cash raised at Series A stage.⁶ The gap between rounds did not get solved. It got financed, and the founder paid for it in dilution.

Signaling, the pricing lens

Assumes: becoming default alive is expensive enough that weak operators will not do it.

Fits because: strategic actors with asymmetric information scored 0.7.

Breaks when: capital is abundant in your specific category. Then growth is rewarded and efficiency is read as low ambition.

Counteracts: optimising the narrative instead of the cash.

May reinforce: premature austerity, and cutting the thing that was actually working.

GEER: the levers, cheapest and most reversible first

Five channels carry your exposure: growth rate, gross margin, committed cost base, currency mismatch between revenue and costs, and collection lag. Pull the cheap, reversible levers before the expensive, permanent ones.

1. Recompute the number. First working day of every month, three inputs, same method. Cash on hand, net burn averaged over three months, revenue growth averaged over three months. Costs an hour.
2. Collect what you already earned. Days sales outstanding is cash sitting in someone else’s account. No repricing, no headcount decision. Days to weeks.
3. Cut discretionary spend priced in a currency you do not earn. Cloud commitments, tooling, paid acquisition. Reversible inside a month.
4. Move the payment term. Deposit up front, annual prepay at a discount, milestone billing. Weeks, and it improves the growth line rather than shrinking it.
5. Match cost currency to revenue currency. Local suppliers, local-currency contracts, or price in dollars where the customer can bear it. Months, partly irreversible.
6. Cut headcount. Statutorily slow, expensive in the first month, and irreversible in practice. This is the correction lag in person.
7. Raise. Slowest, most dilutive, and the only lever whose availability someone else controls.

No-lever flag: if your outflow is fully committed, meaning debt amortisation plus statutory notice plus lease, you hold no reversible lever and your correction lag exceeds your runway. That is a financing-required state, and no amount of discipline converts it into an operating problem. Write that sentence down in those words and re-plan the quarter around it.

RADAR: the portfolio, dated

DO NOW, by T+7. Reversible, dominant across every scenario.

1. Compute months-to-zero on three inputs. One page. No scenario tabs, no best case.
2. Subtract twelve months of contractually committed outflow from cash on hand. Debt service, notice periods, leases. What remains is your real cash.
3. Restate the number in the currency your costs are actually priced in, not your reporting currency.
4. Measure your correction lag by reading your own contracts. Notice periods, severance, exit clauses. One afternoon.
5. Send the number to your co-founders and your board in a format you will not change again.

HEDGE, by T+28. Cheap insurance against the number turning on you.

1. Build the cut list that reaches default alive, with the statutory notice period written on every line. Do not execute it. Costs a day.
2. Move one month of foreign-currency spend onto a local or usage-based contract.
3. Convert one customer to annual prepay, even at a discount that hurts.

DEFER AND TRIGGER. Irreversible, so wait, and pre-commit the observable trigger now.

1. Defer: the layoff, the pivot, the down round, the debt facility.
2. Trigger to execute the cut list: two consecutive monthly readings where months-to-zero falls below your correction lag plus three. For most teams that lands near six months, which is where Graham’s pinch sits, now derived rather than assumed.
3. Counter-trigger: three consecutive months where revenue growth outruns burn growth. Hold the cut list, redirect the effort to collection and pricing, keep recomputing.

If you are the one pricing the runway. DO NOW: ask for months-to-zero, the three inputs behind it, and the date it was last computed. The date tells you more than the number. HEDGE: write the monthly reading into the reporting covenant. DEFER: the bridge decision until you have three consecutive readings produced by the same method.

CHAIN: what usually happens next

Match the reference class on structure. Any business with a largely fixed cost base, a compounding revenue line and an external financing option whose availability is uncorrelated with its own performance runs this machinery. Venture-backed startups qualify. So do import-dependent distributors and project firms billing government.

Base rate: about half of seed-stage companies reach a Series A within four years, and about one in six within two.^{4, 6} Second-order consequence: the standard 18-month runway plan is mispriced against a 25-month median gap, so companies arrive at the fatal pinch on schedule. Third-order: bridge rounds absorb the mismatch and the dilution lands on the founder.⁶

Subtract the counterfactual before you over-read the decline. Part of the graduation-rate fall is composition. The 2021 and 2022 cohorts were unusually large, which inflates the denominator. Your market got harder. It did not get as much harder as the headline rate implies.

Matrix-break flag, three of them.

Debt. African tech raised 4.1 billion dollars in 2025 across 570 deals, and 1.6 billion of that was debt, up 63% year on year, which is 41% of all capital on the continent.⁹ Rational, in a market where formal small and medium enterprises in developing countries carry a 5.2 trillion dollar annual financing gap.¹⁰ It also breaks the model. Graham’s calculation assumes your cost base is discretionary. You cannot cut your way out of an amortisation schedule. If any part of your capital is debt, run default alive net of debt service and treat the remainder as your only real degree of freedom.

Currency. The official naira rate went from 425.98 to the dollar in 2022 to 1,478.97 in 2024.¹¹ A company earning naira and paying dollars for cloud, tooling and senior contractors lost about 70% of its purchasing power without editing a single line of its budget. Your burn is partly set by a central bank. Recompute in the currency you actually pay.

Concentration. Global venture funding reached 425 billion dollars across more than 24,000 companies in 2025, the third-largest year on record, and five companies took 84 billion of it, a fifth of the total.¹² The headline says funding recovered. Five companies took a fifth of it. Do not update your runway plan on someone else’s market.

What this ensemble cannot see

Four blind spots, and they are load-bearing.

Whether your growth rate is a rate. Every model here treats revenue as a compounding series. Three months of lumpy purchase-order revenue, a harvest cycle or one large public tender is a sequence of events, and extrapolating it produces a confident number that means nothing.

Concentration inside the revenue. A burn model cannot see that 60% of your line comes from one distributor whose procurement head is leaving. Months-to-zero looks identical the day before that call and the day after.

Whether you can raise. Fund cycle position, mandate drift, an LP conversation last Tuesday. Invisible from outside, and routinely mistaken for a verdict on the business.

Your own honesty. The output is only as good as the burn number, and the burn number is assembled by the person with the most to lose from what it says. Cooper, Woo and Dunkelberg surveyed 2,994 entrepreneurs: 81% put their odds of success at 70% or better, and 33% put them at 100%.¹³ Treat that optimism as a measurement error with a known sign. Correct for it deliberately, or it corrects your runway for you.

So stop forecasting what you cannot see and go compute what you can. By T+7, produce months-to-zero on three inputs, in the currency your costs are priced in, net of committed outflow. Write your correction lag next to it. Put a recurring entry on the first working day of every month and recompute. When the number drops inside your correction lag plus three, execute the cut list you built at T+28. Everything else on the page is a forecast. That number is a fact, and it is the only one you control.

Sources and notes

1. Graham, P. “Default Alive or Default Dead?” October 2015. Full essay. Source of the default alive definition, the fatal pinch as “default dead + slow growth + not enough time to fix it”, the observation that half the founders he spoke to did not know their status, the “eight or nine months” starting point, and the warning that hiring too fast is the largest killer of funded startups. Also mirrored in the Y Combinator library.
2. Granovetter, M. “Threshold Models of Collective Behavior.” American Journal of Sociology 83(6), 1978, 1420-1443. Full text.
3. Employment Act, 2007 (Kenya), section 40, “Termination on account of redundancy.” Requires notice to the employee and the labour officer not less than one month before the intended date, one month’s notice or wages in lieu, and severance of not less than fifteen days’ pay per completed year of service. Consolidated text, Kenya Law.
4. Carta, Winter 2025 State of Seed report, covering roughly 50,000 startups. Median seed-to-Series-A time of 2.1 years against 1.5 years in 2019, 49.1% graduation for the 2019 first-quarter cohort by the sixteenth quarter, median seed raise of about US$4M at a US$20M post-money valuation. Figures as reported and summarised at SaaStr.
5. Carta data on seed graduation, reporting 30.6% of first-quarter 2018 seed companies reaching Series A within two years against roughly 15% for the 2022 cohort, as compiled by Chronograph, October 2024.
6. Carta State of Private Markets data on bridge financing, second quarter 2025: bridge rounds at roughly 16.6% of all capital raised, from 11.8% a year earlier and under 10% across 2021, and 22.5% of all cash raised at Series A stage; median seed-to-Series-A duration of roughly 616 to 696 days; 15.5% two-year conversion for the 2023 seed cohort. Compiled at SeedScope.
7. Spence, M. “Job Market Signaling.” Quarterly Journal of Economics 87(3), 1973, 355-374. Full text.
8. Sacks, D. “The Burn Multiple.” Craft Ventures, 23 April 2020. Burn Multiple = Net Burn / Net New ARR, presented as a way to judge whether burn is too high “in any given month, quarter, or year.” Full post.
9. Partech, 2025 Africa Tech Venture Capital Report, published 22 January 2026. US$4.1B total across 570 deals, of which US$1.6B was debt (up 63% year on year, 107 transactions, 41% of total capital), US$2.4B equity across 462 deals, 72% of capital to Kenya, South Africa, Egypt and Nigeria, and seed-stage investor participation down 7% year on year. Announcement and report page.
10. IFC, World Bank, SME Finance Forum and IMF, MSME Finance Gap study: 65 million enterprises, or 40% of formal micro, small and medium enterprises in developing countries, hold an unmet financing need of US$5.2 trillion a year. Data site.
11. World Bank, official exchange rate indicator PA.NUS.FCRF, Nigeria: 425.98 NGN per US$ in 2022, 645.19 in 2023, 1,478.97 in 2024, 1,518.38 in 2025 (period averages). Indicator page; machine-readable series at the World Bank API.
12. Teare, G. “Global Venture Funding In 2025.” Crunchbase News, 7 January 2026. US$425B into more than 24,000 companies, up 30% on 2024, third-largest year on record; OpenAI, Scale AI, Anthropic, Project Prometheus and xAI each raised more than US$5B and together took US$84B, about 20% of the global total; AI-related companies took US$211B. Report.
13. Cooper, A. C., Woo, C. Y., and Dunkelberg, W. C. “Entrepreneurs’ Perceived Chances for Success.” Journal of Business Venturing 3(2), 1988, 97-108. Record. The sample of 2,994 entrepreneurs, the 81% who rated their chances at 70% or better and the 33% who rated them at 100% are reported in Salamouris, I. S., “How Overconfidence Influences Entrepreneurship,” Journal of Innovation and Entrepreneurship 2:8, 2013, open access.