<- all tokdocs

CD3 Prioritization: Cost of Delay Divided by Duration, Explained

By , founder of Mavric and Apso

View on TikTok ->

CD3 stands for Cost of Delay Divided by Duration, a way to order a backlog so the most valuable work per unit of time ships first. A TikTok from @_vibe_academy states the formula correctly and suggests pasting your backlog into an AI tool to compute CD3 scores. The formula is right, but the inputs it recommends ("rough effort tags") use effort where CD3 needs calendar duration, and they include no cost of delay at all.

What is CD3?

Black Swan Farming, where Joshua J. Arnold publishes his Cost of Delay material, defines CD3 as "a prioritisation/scheduling method that maximises the value delivered in a given time period when you have limited capacity." The rule is simple: "the priority order of features ... is determined by dividing the estimated Cost of Delay by some estimate of duration: the higher the resulting score, the higher the priority."

The video says the same thing: "Cost of delay divided by duration. That's CD3 and it turns gut feel into a number you can defend."

CD3 worked example

The video gives no numbers, but the Black Swan Farming CD3 page has a worked example with three features:

  • Feature A: $1,000 per week cost of delay, 5 weeks duration, CD3 score of 200
  • Feature B: $4,000 per week, 1 week, CD3 score of 4,000
  • Feature C: $5,000 per week, 2 weeks, CD3 score of 2,500

Running them first-in-first-out (A, B, C) accrues $69,000 of total delay cost. Running them in CD3 order (B, C, A) accrues $27,000, a 61 percent reduction.

CD3 vs WSJF: what is the difference?

They use the same arithmetic. Donald G. Reinertsen set out Weighted Shortest Job First as cost of delay divided by duration in The Principles of Product Development Flow (Celeritas Publishing, 2009). Black Swan Farming describes CD3 as "one specific form of the" WSJF (Weighted Shortest Job First) "queuing method." CD3 is Arnold's practitioner label for Reinertsen's rule, not a rival framework.

The Scaled Agile Framework's version is the one that differs. SAFe's WSJF guidance says "WSJF is estimated as the relative cost of delay divided by the relative job duration," and lists four factors: "relative user and business value, time criticality, risk reduction and/or opportunity enablement, and job size." SAFe builds its numerator by adding relative scores instead of estimating a cost of delay in money per unit of time. A secondary explainer at agility-at-scale says these are scored on a modified Fibonacci scale of 1, 2, 3, 5, 8, 13, 20.

If you ask an AI for "CD3 scores" and get back a sum of 1-to-10 value, urgency and risk ratings over story points, you have SAFe-style WSJF under a CD3 label.

Is duration the same as effort in CD3?

No, and the difference changes the ranking. The video says, "So you paste your backlog, ticket titles, rough effort tags, any revenue or support cost signals you have, and ask AI to compute CD3 scores."

The agility-at-scale explainer puts the distinction this way: "The denominator is job duration, a proxy for how long the work takes relative to other items in the backlog, not effort or cost." Its example is a six-person team working two weeks, which is twelve person-weeks of effort and two weeks of duration. Duration matters because it is the window during which the cost of delay keeps accruing on everything queued behind the item.

The gap is largest in two cases:

  • A one-day change blocked for three weeks on a vendor approval is low effort and long duration. Scoring it on effort inflates its CD3.
  • A large build split across eight people is high effort and short duration. Scoring it on effort deflates its CD3.

Black Swan Farming's at-scale guidance treats duration as elapsed time and uses the longest duration among the teams working on an option.

What counts as cost of delay?

Black Swan Farming defines Cost of Delay as "the partial derivative of the total expected value with respect to time" and credits the idea to Reinertsen, quoting him: "Cost of Delay is the golden key that unlocks many doors." In practice it is a rate, such as dollars per week, which is why dividing by a duration produces a meaningful score.

The video's inputs contain no such rate. Ticket titles carry no economics, and effort tags belong in the denominator. Revenue or support cost signals could feed a numerator, but the phrase "any ... you have" concedes they may be missing. An AI working from that list is guessing the cost of delay. The video also never shows an AI tool, a prompt or any scored output.

For comparison, Arnold and Özlem Yüce's Maersk Line experience report for Agile 2013 estimated delay costs in money per week, roughly $210,000 per week on one sample feature. It reported lead time cut from a 150-day average to 108 days on one programme and 60 on another.

Can you defend a CD3 score?

You can defend the order, not the size of the gap. Black Swan Farming's page on proportionality warns that "the difference in outcomes isn't proportional to the CD3 scores" when options ship in sequence. Its conclusion: "You should still prioritise the option that has the highest CD3 though, as this will accelerate your desired outcome the most."

A score of 4,000 against 2,000 supports doing the first item first. It does not show the first item is worth twice as much. The video's line that "What you get back is a ranked shortlist, not a finalist" fits that reading.

Bottom line

The video's formula is correct, and CD3 is a specific form of Reinertsen's WSJF. Before asking an AI to rank your backlog, give it a cost of delay per week and a calendar duration for each item, not effort tags. Treat the output as an order to discuss, since the scores themselves are not proportional to value.

Sources

Published October 2, 2026. Based on a TikTok by @_vibe_academy. Full video notes (markdown)