Topic 50 of 76 · Software engineering and digital delivery

Cost of Delay (CoD)

Cost of Delay is the economic value lost per unit time that a feature, product, or service is not delivered. It is the single strongest bridge between software delivery metrics and health economics: it converts "we shipped late" into currency — or into QALYs.

Why it matters

Reinertsen's rule: "If you only quantify one thing, quantify the Cost of Delay." Most organizations know what a project costs but not what a month of delay costs, so they optimize budgets while hemorrhaging time-value. For healthcare software the stakes are literal: every week a pathway improvement is delayed, patients wait longer in worse health states. CoD is the strongest mathematical framework to present to NHS stakeholders because it prices the absence of your software.

The math

CoD = benefit per unit time forgone while undelivered   (£/week or QALYs/week)

Total delay loss = CoD × delay duration

For prioritization, see wsjf-and-cd3.md: CD3 = CoD / duration.

For clinical software, denominate in health as well as money:

CoD_health = patients affected per week × QALY gain per patient
CoD_money  = CoD_health × λ (willingness-to-pay threshold, £20k–30k/QALY)
             + operational savings per week forgone

Worked example

Operational: software saves £200 per patient on a pathway; a trust processes 50 such patients/week.

CoD = 200 × 50 = £10,000/week
A 10-week procurement delay costs 200 × 50 × 10 = £100,000 in avoidable waste.

Clinical: a triage improvement removes 5 weeks of waiting (utility 0.68 → 0.80 sooner) for 100 patients/week:

QALY gain per patient = (5/52) × 0.12 ≈ 0.0115
CoD_health = 100 × 0.0115 = 1.15 QALYs/week
CoD_money  = 1.15 × £20,000 ≈ £23,000/week of health value

A 6-month deployment delay "costs" ~30 QALYs — the argument that reframes an IT go-live slip as a clinical event. (Benchmark for scale: Black Swan Farming's famous Maersk analysis found single features with CoD ≈ $200k/week that had waited 38 weeks.)

Software engineering connection

CoD is the metric that makes DORA lead time and flow efficiency financially legible: lead time × CoD = money (or health) burned in queues. Uses:

  • Prioritization: rank work by CoD/duration (WSJF/CD3) instead of loudest-stakeholder.
  • Process economics: a 2-week release cadence has an expected delay cost of ~1 week × CoD per feature versus continuous delivery — price the batch.
  • Procurement: NHS procurement cycles of 6–18 months have a CoD; showing it changes urgency conversations (see budget impact analysis for the affordability counterpart).

Pitfalls

  • Assuming linear CoD: some work has deadline-shaped value (regulatory dates — infinite CoD after the date, zero before) or decaying value (first-mover windows). Classify the urgency profile before multiplying.
  • CoD on outputs nobody wants: delay only costs if the thing has value; garbage delayed is free.
  • Double counting delay and discounting: discounting already prices time on multi-year horizons; CoD is the within-horizon operational version. Use CoD for weeks/months, NPV shift for years.

Sources