Cost of Delay

Cost of delay is the economic consequence of doing work later rather than now: value received later or lost, expense continuing for longer, or additional risk exposure. It makes the consequences of time explicit in prioritization. Two items with equal estimated benefit and effort may deserve different timing if one faces a deadline or continuing loss while the other can wait. Specify the period and how the consequence changes over time; a constant weekly rate is a modeling assumption, not a property of every delay.

What it is made of

SAFe uses a relative cost-of-delay score in weighted shortest job first (WSJF), combining user and business value, time criticality, and risk reduction or opportunity enablement. The components help compare items: a deadline may change time criticality, and an enabling capability may matter because other work depends on it. This relative scoring model is distinct from a currency-valued delay-cost curve. Do not convert an arbitrary relative score directly into money or assume that its scale measures a cash loss per week.

For an internal platform, separate elapsed waiting from avoidable labor, blocked outcomes, and incremental workaround expense. If eight teams each have one request and each waits three weeks, the cohort contributes 8 × 3 = 24 team-weeks of aggregate elapsed waiting. No monthly rate follows without a request frequency and observation period. Nor does the total establish idle labor: teams may work on other tasks. Measure avoidable hands-on work, what delivery was actually blocked, and how substitute cloud costs differ from the proposed service. Do not charge both all waiting time as lost labor and the same delayed output as an independent loss. Opportunity costs need explicit assumptions about what would otherwise have happened.

Turning it into an order

SAFe WSJF divides relative cost of delay by relative job duration. Effort can proxy duration only under suitable staffing and capacity assumptions. For example, with constant delay-cost rates of 100 and 60 units per week and durations of ten and one weeks, the ratios are 10 and 60, favoring the second job under the simple sequential model. Dividing instead by ten and twenty person-weeks of effort would give 10 and 3 and reverse that order. Intercom RICE uses a different denominator: person-months of effort, with reach × impact × confidence in the numerator. Keep lifecycle economics separate: three person-months to build plus half a person-month per operating month requires a horizon before addition. Neither ratio alone resolves dependencies, mandatory deadlines, uncertain outcomes, or shared capacity. Revisit priorities when those inputs change.

Past expenditure is sunk and gives a project no entitlement to priority. Work already completed can nevertheless shorten its remaining duration. At the same delay-cost rate of 100 units per week, remaining durations of two and five weeks give ratios of 50 and 20. Use the work and duration still required, and account for switching or restart costs and dependencies. Ignoring sunk costs does not mean ignoring the project’s current state.

Use estimates with units, ranges, and an explanation of their uncertainty. A relative score is not inherently reliable just because its scale is small, and a currency estimate should not imply more precision than its evidence supports. Distinguish a supported estimate of zero cost from an unknown cost, a plausible range, or a case where this model does not apply. A request without a problem statement is not evidence of zero value or zero urgency; decide whether a bounded investigation is worthwhile. The Vote Always Went to the Forty explains how this fits with platform scope, segments, and mandatory gates. Weighted scoring provides a general comparison method. The underlying value hypothesis needs evidence capable of separating the proposed effect from other causes: measuring an outcome afterward does not, by itself, establish attribution.

References:

• Scaled Agile, Weighted Shortest Job First, SAFe Knowledge Base, publicly visible definition.
• Sean McBride, RICE: Simple prioritization for product managers, Intercom.