When the cost of late error discovery exceeds 5x early…

When This Fires

• At every phase boundary of a multi-phase project (design → build → test → deploy)
• Before committing to the next phase of any staged investment
• When deciding how much testing, review, or validation to invest at each checkpoint
• Complements Test your riskiest assumptions first and cheapest — list… (assumption testing) with ongoing phase-gate detection

Common Failure Mode

Rushing through phase gates because "we're behind schedule": "Let's skip the design review and go straight to implementation." If the cost ratio is 10:1 (design errors caught now cost $10K, caught in production cost $100K), skipping the review saves days but risks $90K in unnecessary rework. The schedule pressure is real but the cost asymmetry is larger.

The Protocol

(1) At each phase boundary, estimate two costs: Cost of finding an error now: what does it cost to discover and fix a foundational error at this stage? Cost of finding the same error after the next phase: what would it cost if the error survived undetected through the next phase? (2) Calculate the ratio: late cost ÷ early cost. (3) If ratio < 5:1 → proceed normally. Detection investment at this stage has low leverage. (4) If ratio ≥ 5:1 → increase detection before proceeding: additional reviews, prototypes, tests, expert consultations, or validation experiments. The detection investment is justified by the cost asymmetry. (5) The higher the ratio, the more detection to invest. A 50:1 ratio justifies pausing for weeks if necessary. A 5:1 ratio justifies a few extra days of verification.