Failure

How do companies actually die?

1. 01Aurygine synthesisEvidence quality: medium

   Channel concentration: when one working channel becomes the risk

   Aurygine's own synthesis across observed practice. No single work is reproduced.

   Extracted pattern

   • A channel you do not own can be repriced or removed without notice.
   • Diversifying before you have one channel working is procrastination; diversifying after is insurance.
   • The second channel is nearly always slower and worse than the first, and should be judged on payback, not volume.

   Decision pattern

   • Situation. One channel produces most of the company's new customers.
     Move. Keep investing in the working channel, but run one small, funded test in a structurally different channel each quarter.
     Trade-off. The test costs money and attention that the working channel would convert better today.
     Breaks when. The company has not yet found a single channel that works — then focus beats diversification.

   Practical implication

   Measure the share of new business each channel produces. Above roughly 70% from one source, fund a deliberate second bet — sized so failure is affordable and success is measurable.

   Counter-pattern / contradiction

   • Local service businesses often survive fine on one referral channel for years, because the channel is a relationship, not a platform.

   Failure modes

   • Assuming a platform's current economics are permanent.
   • Judging the second channel by first-channel efficiency and killing it too early.

   What was measured

   share of new customers by channel · payback period per channel · owned audience size

   Stage relevance

   First customers · Traction · Scaling

   Source

   Aurygine · 2026-02-02
2. 02SourcedEvidence quality: high

   Default alive: knowing whether the company survives without new money

   Summarised in Aurygine's words from a publicly readable source, linked below.

   Extracted pattern

   • Every founder should be able to answer, without a spreadsheet session, whether the company reaches profitability on current cash and current growth.
   • Not knowing the answer is itself the finding.
   • Once the answer is no, the choices narrow to: grow faster, cut cost, or raise — and delay removes options.

   Decision pattern

   • Situation. Spending exceeds revenue and the founder is unsure how serious it is.
     Move. Compute months of runway at current burn, then ask whether current growth closes the gap before the cash runs out.
     Trade-off. Forces an uncomfortable conversation earlier than the founder would choose.
     Breaks when. Revenue is highly seasonal or lumpy, where a single month's burn misreads the year.

   Practical implication

   Keep the runway number visible and honest. The decision that matters is not 'are we doing well' but 'does the company survive on the cash it already has'.

   Counter-pattern / contradiction

   • Deep-tech and infrastructure companies are structurally default-dead for years by design; the measure there is milestone risk, not runway alone.

   Failure modes

   • Treating a funding conversation as a plan.
   • Cutting cost so late that the cut has to be twice as deep.

   What was measured

   months of runway · monthly burn · growth rate · gross margin

   Stage relevance

   First customers · Traction · Scaling · Turnaround

   Source

   Paul Graham · 2015-10-01 original