Cost of Delay: How to Calculate It and Use It to Decide
Cost of delay is the value you lose for each week a project waits. Learn the formula, a worked CD3 example, and how to log the estimate as a bet you can review.

On this page
- What cost of delay actually measures
- The cost of delay formula
- A worked example with CD3
- How to estimate when you have no data
- Where cost of delay goes wrong
- Why a cost of delay estimate is really a bet
- What a good cost of delay record contains
- Perspective: the number is less important than the record
- Putting cost of delay into a review rhythm
- Keeping the record in Betlog
- FAQ
- Sources
Cost of delay is the value you lose, or fail to gain, for every unit of time a piece of work stays undelivered. You state it as a rate, such as dollars per week, and that rate lets you compare projects with numbers instead of volume. To calculate it, estimate what the work is worth, work out how that value changes as time passes, and divide by the time unit you care about. Then write the estimate down, because it is a forecast, and forecasts deserve a record.
TL;DR
- Cost of delay (CoD) is a rate: the value lost per week or month of waiting. It is not a total project budget.
- Basic formula: value at stake divided by the time period it is lost over. To sequence work, divide CoD by duration (CD3).
- Keep labor and project expenses out of the calculation. They are cost, not opportunity value.
- Aim for “right enough to change the order,” not precise to the dollar.
- Log each CoD estimate with a probability and a review date, so you learn how good your estimates are.
What cost of delay actually measures
The Wikipedia entry on cost of delay describes it as a way of communicating the impact of time on the outcomes we hope to achieve. More formally, it is the partial derivative of total expected value with respect to time. In plain terms: if you ship this a week later, how much worse off are you?
That framing matters because most priority debates run on two confused ideas. Black Swan Farming points out that cost of delay combines value and urgency, two things people routinely blur together. A project can be very valuable and not urgent: its value is the same next quarter. Another can be modest in value and brutally urgent: the value disappears if you miss a date. Cost of delay puts both into one number.
Three components are commonly used to estimate it:
- Business value: revenue gained, retention protected, or cost removed once the work ships.
- Time criticality: how fast that value decays. A seasonal launch, a competitor’s move, or a contract deadline makes the decay steep.
- Risk reduction or opportunity enablement: what the work unlocks, or what danger it removes.
The first number you need is the rate. Everything else follows from it.
The cost of delay formula
At its simplest:
Cost of delay = value lost ÷ time period over which it is lost
If a pricing change would add $20,000 of monthly recurring revenue and it ships a month late, the delay costs roughly $20,000 for that month. Expressed as a weekly rate, that is about $4,600 per week. Use the unit that matches your planning cadence. Weekly works for most software teams.
This gets more interesting when the value is not constant. Picture three shapes:
- Linear: value leaks at a steady rate, like a feature customers would pay for starting on day one.
- Deadline: nothing is lost until a date, then most of the value vanishes. Compliance dates and trade shows work this way.
- Decaying window: the opportunity shrinks the longer you wait, like a first-mover position.
Write down which shape you believe applies. It changes the answer more than the dollar figure does.
Keep expenses out of the number
Practitioners at Playbook warn against putting labor or project expenses inside the cost of delay calculation. Expenses are project cost. Cost of delay is opportunity value. Mixing them hides the economic benefit of launching sooner behind the budget of the work itself.
Some teams still want to see the carrying cost of a stalled team, and that is a fair separate question. If six people at $2,000 a week each sit on a four-week slip, that is $48,000 of payroll buying nothing new. Track it next to CoD, not inside it.
A worked example with CD3
Cost of delay alone does not tell you what to do first. A big CoD that takes eight weeks to deliver may lose to a smaller one that takes a week. The scheduling version divides cost of delay by duration, known as CD3. It is a form of Weighted Shortest Job First (WSJF), as the Wikipedia entry explains.
Here is a hypothetical backlog for a 12-person SaaS team. The numbers are invented for illustration.
| Work item | Cost of delay | Duration | CD3 (CoD ÷ weeks) |
|---|---|---|---|
| A. Self-serve onboarding rebuild | $10,000/week | 4 weeks | 2,500 |
| B. Fix billing retry failures | $6,000/week | 1 week | 6,000 |
| C. Enterprise SSO for a pending deal | $20,000/week | 8 weeks | 2,500 |
By raw cost of delay, SSO looks like the winner. By CD3, the billing fix goes first: one week of work stops a $6,000-per-week leak. Then you choose between A and C, which tie. That tie is useful. It tells you the decision hinges on something other than the numbers, perhaps whether the SSO deal has a hard close date (a deadline shape) or whether onboarding improvements compound.
Total cost of waiting also depends on order. Run B, then A, then C and compare it with other sequences. A short spreadsheet will do it. The gap between the best and worst sequence is the real prize.

How to estimate when you have no data
The common objection is that nobody knows the numbers. That is true, and it is not a reason to skip the exercise. The Software Engineering Guide summarizes Don Reinertsen’s argument: the underlying forecasts are never exact, but their relative sizes are often robust enough to show which work earns the most for the time invested.
A practical routine:
- Pick one outcome. Revenue, retention, hours saved, or risk avoided. Avoid blending them in one line.
- Start from a number you already trust. The sales forecast behind the project approval, the current churn rate, hours your team spends on a manual task.
- Estimate per week. For time saved internally, multiply hours saved per week by a loaded hourly rate.
- Choose the decay shape. Linear, deadline, or window.
- Give a range. “$4,000 to $9,000 per week” is more honest than “$6,500.”
- Ask stakeholders to defend it. If everything is priority one, do not fight about volume. Ask each requester to put a weekly figure on their urgency and defend it. That common currency often breaks the “everything is urgent” deadlock.
If you manage a mix of tasks and commitments, a comparison of task prioritization frameworks can help you decide where cost of delay fits next to simpler methods.
Where cost of delay goes wrong
Analysis paralysis. Black Swan Farming notes that the model only needs to be right enough to change the sequence. If a second hour of modeling will not change the order, stop.

False precision. A single figure like $6,512 per week signals certainty that does not exist. Use ranges and say how confident you are.
Counting value twice. If two projects share the same revenue upside, do not credit both in full.
Ignoring the decision’s other side. Shipping early has costs: quality risk, support load, fatigue. Cost of delay measures the price of waiting, not whether to proceed. You still weigh trade-offs.
Treating it as a one-time report. Estimates go stale. When a competitor launches or a deal slips, the decay shape changes and so does the order. Revisit the number whenever the situation changes.
Treating a delay as proof. Later, you may see that waiting cost little, or a lot. Judging the original estimate by that single result is the trap Annie Duke calls resulting. A well-reasoned call to wait can still lose.
Why a cost of delay estimate is really a bet
Every cost of delay number is a prediction about the future: a forecast of value, a guess at how fast it decays, and an assumption about what competitors do. It is a bet with incomplete information. That has two practical consequences.
First, you should attach a probability. “We are about 65% confident this feature is worth at least $5,000 a week” is a statement you can check later. “This is really important” is not.
Second, you should write it down before the outcome, with what would prove it wrong. Otherwise, when the launch lands, memory will quietly rewrite what you expected. Our piece on making decisions with incomplete information walks through the same discipline for founders.
The same logic applies when waiting itself is the decision. Choosing to delay for more data has a price per week. Writing that price next to the decision makes the trade explicit. A short premortem before committing to a sequence also surfaces the ways your CoD estimate might be wrong.
What a good cost of delay record contains
Whether you use a spreadsheet or something purpose-built, a useful entry has these fields:
- The decision: which item goes first, and what you are deferring.
- CoD estimate: a weekly range, the metric it is measured in, and the source of the base number.
- Decay shape: linear, deadline, or window, with the date if there is one.
- Confidence: a probability that the estimate lands within your range.
- Trade-offs accepted: what the deferred work will cost while it waits.
- Review date and deciding metric: when you will compare the estimate with reality.
- What would prove you wrong: for example, “if onboarding conversion does not move by 3 points in six weeks, our value estimate was too high.”
Our decision log template gives you a copyable version of this structure in about ten minutes.
Perspective: the number is less important than the record
We built Betlog around one idea from Annie Duke’s Thinking in Bets: every decision is a bet, and its outcome depends on decision quality and luck. Cost of delay is a good example of why that matters.
Teams that adopt cost of delay usually learn something in the first meeting: nobody agrees on the value of the work, and nobody had said so. That disagreement is the benefit. Surfacing it is worth more than the arithmetic.
What tends to go missing is the follow-up. A team ranks work by CoD, ships in that order, and never revisits the estimates. Six months later nobody can say whether the “$10,000 per week” claims were any good. Without that feedback, each new estimate is as uncertain as the first.
So we suggest treating each estimate as a bet with a stated probability, then reviewing it. After a dozen bets you can see whether your 70% calls land about 70% of the time. If your “high urgency” items keep turning out to be mild, you have learned something no formula could tell you. Our decision audit is one way to run that review so luck does not get credited as skill.
— Cesar
Putting cost of delay into a review rhythm
A lightweight cadence keeps the model alive:
- At prioritization: write the CoD range, decay shape and confidence for each candidate.
- At each status call: ask one question: has anything changed that moves the rate or the shape? Update only if so.
- At the gate: before moving an item from Prioritized to Running, check the CD3 order against the current numbers. Our 90-minute decision gates approach covers how to run this without the meeting dragging.
- After shipping: compare the realized value with the estimate and note how much was skill and how much was luck. The postmortem questions we use are a starting list.
Speed matters in some contexts more than others. In lead-driven service businesses, the value of a response decays within minutes, which is why material on speed to lead is a useful illustration of a steeply decaying window. Your product work probably decays more slowly, but the question is the same: how fast does the value drain?
Keeping the record in Betlog
Betlog is a decision journal for founders and small teams. Each bet captures the hypothesis, the expected outcome, your confidence as a probability, the trade-offs, the metrics that decide it and what would prove it wrong. Bets move through Idea, Prioritized, Running, Reviewing and Decided, and close as Won, Killed or Inconclusive with a post-mortem that separates skill from luck. A cost of delay estimate fits naturally as part of the hypothesis and the deciding metric.
There is one plan: $39 per month or $390 per year. You can see what is included on the pricing page, or start at betlog.sh.
FAQ
How do you calculate the cost of delay?
Estimate the value the work delivers, decide how that value changes over time, and express the loss as a rate, such as dollars per week. For sequencing, divide that rate by the work’s duration to get CD3, then do the highest score first. Use ranges rather than a single figure.
Can cost of delay really be quantified?
Only roughly, and that is enough. The inputs are forecasts, so the absolute figure will be wrong to some degree. What matters is whether the relative sizes are good enough to change the order of work. Log your estimates and compare them with outcomes to see how well you do.
What is a good example of cost of delay?
If a billing fix would stop $6,000 per week of failed renewals, every week it waits costs $6,000. A one-week fix therefore beats an eight-week project with a higher weekly figure but a lower CD3. The table above shows the full comparison with invented numbers.
Should salaries be included in the calculation?
No. Salaries and other project expenses are cost, while cost of delay measures opportunity value. Track team carrying cost separately if you want to see it. Keeping them apart avoids hiding the benefit of shipping sooner.
Sources
- Cost of delay, Wikipedia
- Cost of Delay, Black Swan Farming
- Cost of Delay Project Model Examples, Playbook
- 11.6 Value stream mapping and cost of delay, Software Engineering Guide


