CompanyGraph
How small changes can have large effects when capacity, liquidity, adoption, or solvency crosses a condition that changes the system’s response.
A Threshold Is a Mechanism, Not a Magic Number
In a linear story, a little more capacity or market share produces a little more result. In a threshold story, the result changes when a condition is crossed. A marketplace may become useful only when enough buyers and sellers can find a match. A factory may cover fixed cost only after enough units pass through it. A lender may stop funding a company when liquidity falls below a covenant or confidence boundary.
The critical level is not universal. It depends on geography, product, customer density, contracts, time, and the response of other participants. A threshold can also be gradual rather than a single point. The useful question is what mechanism changes near the boundary, not whether an analyst can name an exact percentage.
Three Common Sources of Nonlinearity
Positive network feedback occurs when more participants increase the value of joining. A buyer platform with too few sellers offers poor selection; enough liquidity can improve matching and attract more users. The same loop works in reverse when participants leave. Granovetter’s threshold model provides a formal way to describe how heterogeneous participation thresholds can produce cascades without assuming one universal tipping point. Granovetter’s paper is provenance for the mechanism, not a forecast for any platform.
Fixed-cost scale creates a different boundary. A plant can have capacity available while its volume is too low to cover maintenance, labour, depreciation, and compliance. Higher output may reduce unit cost until congestion, overtime, quality loss, or a new expansion requirement reverses the benefit. Minimum efficient scale is therefore an interval tied to a product and process, not a permanent size advantage.
Financial or operating cliffs arise when a covenant, cash balance, service level, or safety margin is crossed. Missing a payment can trigger acceleration. A full server or warehouse can cause waiting times to rise sharply. These thresholds can be contractual or physical, and they often interact.
Why Feedback Can Make the Change Abrupt
Positive feedback amplifies a lead: more participants improve service, which attracts more participants. Negative feedback can stabilise a system or create a ceiling: congestion makes the service worse, reducing demand. A platform can have both loops at once. Growth before a bottleneck and growth after a bottleneck are not the same process.
Threshold effects can be reversible. A network that loses liquidity may recover through subsidies, a new use case, or a complementary service. A plant below efficient scale may improve economics through automation or a contract, while a heavily indebted business may cross a point from which refinancing is impossible. Avoid describing every threshold as an irreversible winner-take-most event.
What Investors Can Observe
- State variable: match rate, utilisation, cash runway, delivery time, defect rate, or another defined measure.
- Response: whether an incremental change now produces a different customer, cost, or financing outcome.
- Feedback: the path by which the result attracts or repels the next participant or resource.
- Buffer: spare capacity, liquidity, inventory, or contractual room before the boundary is reached.
- Countermeasure: what can be funded and authorised before the response changes.
Do Not Confuse a Threshold With Correlation
A sudden market-share change may follow a threshold, but it may also reflect a price cut, regulation, acquisition, supply failure, or measurement change. A high utilisation rate may coincide with better margins because demand is strong rather than because a fixed-cost threshold was crossed. The investor should compare similar units over time and test alternative explanations.
Threshold analysis is valuable when it makes the next unit of capacity, liquidity, or participation intelligible. It becomes storytelling when an exact tipping point is asserted without identifying the underlying mechanism, measurement, and reversible alternatives.