The Failure Library

Why a failure library

Most investment writing collects winners. It studies the businesses that compounded for decades and works backward to find the traits that predicted that outcome. This produces useful pattern recognition, but it also produces survivorship bias — the sample is selected by the outcome being analyzed.

The Failure Library works from the other direction. It collects the base rates investors prefer to ignore: the ways high-ROIC, famous, apparently durable companies become permanent capital mistakes. Bessembinder's research shows that the median stock underperforms cash over its full life. The failure patterns are not rare exceptions; they are the central tendency. The exceptions are the compounders.

A serious long-duration process needs explicit failure pattern recognition alongside winner pattern recognition. If you only study compounders, you learn how to identify the rare right tail. If you also study failures, you learn how to avoid the fat left tail. The Failure Library is the systematic attempt to make the left tail explicit.

The 16 failure patterns

1. Substrate obsolescence. The business is built on a physical, regulatory, or technological substrate that is being replaced by something better and cheaper. The moat is real but temporary — it defends a position that is being made irrelevant. Classic examples include film photography, physical media distribution, and landline telephony.

2. Payer / price-setter compression. The business's economic returns depend on a single or concentrated payer who gradually extracts more of the value. Healthcare services businesses facing insurer consolidation, suppliers facing retailer consolidation, and professional services firms facing procurement rationalization all follow this pattern.

3. Reinvestment-runway exhaustion. The business is excellent but has run out of places to deploy capital at attractive incremental returns. It becomes a dividend machine rather than a compounder — not a failure in the traditional sense, but a permanent derating relative to a business that still has runway.

4. Capital-allocation culture destruction. The founder or capital allocator who built the business's discipline is replaced by operators who optimize for scale or reported earnings rather than returns on incremental capital. The business's culture of capital discipline erodes gradually and then collapses.

5. Balance-sheet time bomb. The business carries debt that is manageable in normal conditions but fatal in a stress scenario. The leverage is often obscured by strong EBITDA coverage ratios that ignore the cyclicality or capital intensity of the underlying business.

6. Roll-up accounting optics. The business uses acquisitions to generate reported EBITDA growth that does not reflect underlying economic returns. Goodwill accumulates, organic growth decelerates, and the gap between reported earnings and free cash flow widens until it can no longer be ignored.

7. Cyclicality mistaken for secular growth. A long upcycle in an end market is interpreted as evidence of structural competitive advantage. The business is valued as a compounder during the upcycle and rerates sharply when the cycle turns. Commodity producers, housing-adjacent businesses, and capital-equipment companies are recurring candidates.

8. Brand without pricing mechanism. The business has high consumer recognition but lacks a structural mechanism that translates brand into durable pricing power. When a lower-priced competitor with comparable quality enters, the brand premium evaporates more quickly than anticipated.

9. Culture destroyed by leadership transition. The founder or transformative leader leaves and is replaced by a manager who lacks the conviction to maintain the cultural practices — decentralization, long-term incentives, candid communication — that produced the prior results. The business degrades gradually over years before the cause becomes visible.

10. Category commoditization. A product or service that was differentiated becomes a commodity as production technology diffuses, IP protection expires, or a better-capitalized competitor decides to compete on price. The moat was real but not permanent.

11. Network-effect reversal. A business with genuine network effects loses critical mass in a key segment, and the network begins to shrink rather than grow. The same dynamics that made the network valuable on the way up accelerate its decline on the way down.

12. Accounting / governance blowup. The business's reported economics were not real. Revenue was recognized too early, costs were capitalized rather than expensed, or related-party transactions obscured the true economics. The blowup is typically preceded by years of clean audits and credible-sounding explanations for the unusual accounting treatment.

13. Terminal-value illusion. The business is valued as if its current economics will continue indefinitely, when in fact the current economics are the peak of a competitive position that is already beginning to erode. The terminal multiple reflects the good years, not the regression to mean that follows.

14. Platform-tax compression. The business generates economics by sitting between two parties and extracting a toll. A more powerful platform — typically a large technology company with a captive user base — enters the same position and charges a lower toll, compressing the original platform's economics without eliminating the market.

15. Founder-dependence reversal. The business's competitive position is inseparable from the specific judgment, relationships, or reputation of one person. When that person leaves, the business loses capabilities that cannot be transferred or institutionalized. The succession risk was underestimated because the founder's contribution was hard to separate from the business's structural position.

16. Demographic / policy tailwind reversal. The business benefited from a multi-decade demographic or policy tailwind that was mistaken for competitive advantage. When the tailwind reverses or plateaus, the underlying economics are revealed to be weaker than the historical returns suggested.