Slip Cost per Week

A sponsor says the launch must hit the date and the architecture must be cut to fit. The architect says the shortcuts will cost more later. Both are asserting and the louder person wins. Neither has computed what a week is worth, which is the number that would settle it.

Slip cost per week takes three inputs: the annual benefit the launch unlocks, the share permanently lost rather than deferred by a delay, and whether a date exists after which the value collapses. The second input is where the argument lives and is almost never stated.

Why it matters

Deliberate technical debt is a loan, judged against what it buys. If a week is worth £7,700 and the shortcut costs three engineer-months to repay, the loan is a bad one — a sentence that ends a month of debate.

The reverse is more expensive. When a launch must land before a peak carrying 40% of annual volume, the slip deletes benefit rather than deferring it, and a week in the wrong month can be worth ten times a week in the right one.

Implementation patterns

• Compute two rates: the deferred rate (annual benefit ÷ 52) and the destroyed rate for a dated window.
• Classify the benefit as deferred (a subscription starting later), destroyed (a seasonal peak, a tender deadline) or competitive (a rival takes the share permanently). Only the first is recoverable by shipping later.
• Put slip cost next to debt cost in one decision record: shortcut, weekly slip avoided, repayment cost, repayment date, owner.
• Decompose the lead time before accepting the date, because compressing delay usually means attacking waiting — an approval queue, supplier onboarding, a shared backlog — not building.
• State the zero case. With a deferred benefit and no date, slip cost approaches the cost of capital, which is a weak case for corners.

Industry example

A general merchandise retailer plans a new category worth £400k of gross margin a year, so a deferred week is about £7,700 — small enough that engineering's two-quarter estimate is affordable. But demand is concentrated: roughly 40% of annual volume falls in a six-week window before the winter holidays.

Missing that window removes about £160k unrecoverable until next year, so a week lost in October costs on the order of £27k and a week lost in February costs £7,700. The decision follows: spend to hit the window, run the back office manually behind a working customer front end, and date the repayment for January. The failure variant is the same business discovering in November that the software shipped and returns, settlement and support were not ready. Slip cost is measured against the whole value stream, not the deployment.

Failure scenarios

• Asserted magnitude. "The cost of delay is huge" is never computed, and the team takes on compounding debt for a benefit that was deferred anyway.
• Deferred treated as destroyed, which funds permanent damage to buy back recoverable weeks.
• Destroyed treated as deferred: the team builds it properly, misses the window, and the business case evaporates while the architecture is admired.
• Debt with no repayment date, so the loan becomes permanent and the next launch pays interest.
• Compressing the wrong step, spending on build capacity when the delay is an approval queue. Adding people late to a waiting-dominated process makes it worse, as the 1975 software-schedule literature documented.

Trade-offs

Computing the number costs a day or two with finance, plus a willingness to accept an answer that undermines your own position. An architect who computes slip cost will sometimes find the shortcut justified, and must then take it.

The estimate carries real uncertainty, mostly in the forecast benefit. Present a range and name the dominant assumption. The shape changes the decision; the magnitude only sizes it.

When not to use it

When the deadline is external. A regulatory date is a constraint, not a trade, and pricing it weekly distracts from descoping.

When the benefit cannot be forecast. For a new market with no comparable, a fabricated number is worse than none; decide on option value instead.

When the shortcut is a correctness or security compromise. Those are not loans, because the downside is unbounded and cannot be netted against a weekly benefit.

Interview question

Q: Your sponsor wants a launch pulled in by six weeks and will pay £150k in contractors for it. Engineering says the shortcuts cost a quarter of rework. How do you decide, and what do you need to know first?

What a strong answer covers: computing slip cost from annual benefit, the deferred-versus-destroyed split and any hard window; comparing it to the £150k and the rework estimate; checking whether six weeks are buyable with people at all, since a waiting-dominated lead time gets worse when staffed harder; a partial answer with a dated repayment; and excluding correctness and security shortcuts.

Quick check

Quiz: A launch worth £400k of annual margin slips a week. What is the cost, and what fact multiplies it tenfold? About £7,700 if the benefit is deferred; a dated window carrying 40% of annual volume destroys rather than defers it.

Flashcard: What three inputs give slip cost per week, and which changes the decision? — Annual benefit, the destroyed-versus-deferred share, and whether a hard window exists. The destroyed share decides; the magnitude only sizes it.