Skip to content
PMPCAPM

Weighted Shortest Job First (WSJF)

Weighted Shortest Job First (WSJF) is a prioritization model, popularized by SAFe, that ranks work by cost of delay divided by job duration (usually estimated as job size), so the items that lose the most value per unit of time spent waiting are done first.

Share:

Know it well enough for the exam? Questions on Weighted Shortest Job First (WSJF) are scenario-based, not definitional.

Practice it free

Explanation

WSJF is used when a backlog holds many competing items of different sizes and a team, a group of teams or a portfolio has to decide the order. In SAFe it is applied to features before PI planning and to epics in the portfolio backlog. It is just as useful outside SAFe whenever "which is most valuable?" debates go in circles, because it forces the group to separate value, urgency, risk and size.

The logic is economic. While one job is being worked on, every job waiting behind it keeps losing value. Doing short jobs with a high cost of delay first minimizes the total value lost across the queue. That only holds when jobs compete for the same capacity; independent work can run in parallel and does not need ranking against each other.

WSJF has limits. The scores are relative judgments, so the ranking is only as good as the conversation that produced them. Dependencies can force a different order, near-equal scores are effectively ties, and the scores go stale as deadlines approach. Treat WSJF as an input to the product owner's decision, re-scored regularly, not as an automatic answer.

WSJF in plain language

Every item in a backlog loses value while it waits. A compliance feature with a legal deadline loses a lot; a nice-to-have dashboard loses very little. That loss per unit of time is the cost of delay. If you can only work on one thing at a time, the order that loses the least value overall is: do the jobs with the highest cost of delay per unit of duration first. That is all WSJF is.

WSJF = Cost of Delay ÷ Job Duration

The idea comes from Don Reinertsen's work on product development flow, where it is also called CD3 (cost of delay divided by duration). The Scaled Agile Framework (SAFe) popularised a lightweight version that teams can score in a meeting without any financial model, and that version is what exam questions mean by WSJF.

How SAFe scores it

SAFe breaks cost of delay into three relative scores and uses job size as a proxy for duration:

ComponentQuestion it answers
User-business valueHow much do users or the business want this? Is there revenue or a penalty involved?
Time criticalityDoes the value decay quickly? Is there a fixed date, a competitor, a customer waiting?
Risk reduction / opportunity enablementDoes doing this reduce a risk, or unlock other valuable work later?
Job size (denominator)How big is it relative to the other items? Stands in for duration.
  1. 1Score one column at a time across all items, not one item across all columns. It keeps the comparisons honest.
  2. 2In each column, give the smallest item a 1 and score the rest relative to it using a modified Fibonacci scale (1, 2, 3, 5, 8, 13, 20).
  3. 3Add the three value columns to get cost of delay.
  4. 4Divide cost of delay by job size. Highest WSJF goes first.

Worked example: four features, one team

A payments platform team has capacity for one feature at a time. The product owner and stakeholders score four candidates:

FeatureBusiness valueTime criticalityRisk / opportunityCost of delayJob sizeWSJF
F1 — Self-service password reset8311234.0
F2 — Analytics dashboard131216131.2
F3 — Audit log for new regulation31352154.2
F4 — API rate limiting121316116.0

The WSJF order is F4 → F3 → F1 → F2. Compare that with what simpler rules would have chosen:

RuleFirst pickWhy it is worse here
Highest business valueF2The biggest job with the lowest urgency blocks everything else for the longest time.
Highest cost of delayF3Reasonable, but ignores that F4 delivers most of its value in a fraction of the time.
Smallest job firstF4Right answer by luck; the rule would also rush trivial items with no value.
WSJFF4Balances value lost per day of waiting against how long each job occupies the team.

F4 wins because it is tiny and removes a large risk (a partner integration that can currently overload the service). F3 and F1 score 4.2 and 4.0: that gap is well inside the noise of relative estimates, so treat them as a tie and let a conversation decide, for example by checking the regulation's effective date. WSJF ranks; it doesn't replace judgment.

Why divide, not multiply?

Doing a short job first delays everything behind it only briefly; doing a long job first delays everything behind it for a long time. Dividing by duration charges each job for the waiting it imposes on the rest of the queue. Multiplying would reward big jobs, the opposite of the intent.

Common exam traps

  • Picking the item with the highest business value.

    Business value is one of three cost-of-delay inputs, and the total is divided by size. A high-value but huge item often ranks low.

  • Putting job size on top of the fraction.

    WSJF = cost of delay ÷ job size. Highest score first. If an answer option multiplies or puts size in the numerator, it is wrong.

  • Treating WSJF as a one-time ranking.

    Time criticality changes as deadlines approach and risks retire. SAFe re-scores before each PI planning event; any team using it should re-score when the backlog is refined.

  • Mixing absolute and relative numbers.

    Scores are relative within one backlog. Comparing a WSJF of 4 from one team to a 4 from another team means nothing, and so does mixing dollar estimates with Fibonacci scores.

  • Assuming the team or the project manager sets priority.

    In PMI’s agile framing the product owner owns backlog ordering, with stakeholder input. WSJF is a tool the product owner uses, not a vote that overrides them.

WSJF vs other prioritization techniques

TechniqueHow it ranksBest when
WSJFCost of delay ÷ size, scored relativelyMany competing items of different sizes; economic sequencing matters
MoSCoWMust / Should / Could / Won’t have this timeAgreeing release scope against a fixed date
Kano modelBasic, performance and delighter featuresUnderstanding customer satisfaction, not sequencing
100-point methodStakeholders split 100 points across itemsSurfacing stakeholder preferences quickly
Simple value rankingHighest value firstItems are all similar in size

Exam questions usually describe a situation and ask which approach fits. If the stem mentions SAFe, cost of delay, or dividing value by size, the answer is WSJF. If it mentions agreeing what is in or out of a release, think MoSCoW. Job size is a relative estimate, normally in the same units the team uses for story points.

Practice and related pages

Key Points

  • •WSJF = Cost of Delay / Job Duration (job size is the usual proxy)
  • •SAFe cost of delay = user-business value + time criticality + risk reduction/opportunity enablement
  • •Scored relatively with a modified Fibonacci scale; the smallest item in each column gets 1
  • •Highest WSJF goes first; close scores are effectively ties
  • •Re-score as deadlines approach and risks change

Exam Tip

If an answer prioritizes purely by business value, it is usually the trap. WSJF divides cost of delay by size, so a small, urgent or risk-reducing item can outrank the most valuable one. The product owner still owns the final backlog order.

Frequently Asked Questions

Related Topics

High-yield topics our learners drill most before exam day.

Burndown Chart

A burndown chart plots the work remaining (usually in story points) against time in a sprint or release, alongside an ideal line to zero, so the team can see each day whether it is on track to finish.

Burn-Up Chart

A burn-up chart plots completed work rising over time against a separate total-scope line, so progress and scope changes are visible independently; the work is finished when the two lines meet.

Relative Estimation

Relative Estimation is an agile technique where work items are sized in comparison to each other rather than in absolute units like hours or days, providing faster and more accurate estimates.

Sprint Review

The Sprint Review is a Scrum event held at the end of the Sprint where the Scrum Team presents the Increment to stakeholders, gathers feedback, and collaborates on what to do next.

Sprint Backlog

The Sprint Backlog is the set of Product Backlog items selected for the Sprint, plus the Sprint Goal and the plan for delivering the Increment.

Timeboxing

Timeboxing is the practice of allocating a fixed, maximum amount of time for an activity, after which the activity stops regardless of whether it is complete.

Resource Leveling

Resource leveling is a resource optimization technique that adjusts activity start and finish dates so that demand for a resource never exceeds its availability, accepting a longer schedule and a changed critical path if necessary.

Deliverables

A deliverable is any unique and verifiable product, result, or capability to perform a service that must be produced to complete a process, phase, or project, such as a software release, an audit report, or a trained support team.

Power Interest Grid (Power/Influence Grid)

The power interest grid is a 2x2 stakeholder mapping tool that plots each stakeholder by power (ability to affect the project) and interest (concern about its outcome) to decide how to engage them: manage closely, keep satisfied, keep informed, or monitor.

Part of

Agile & Hybrid

Study full domain →

Practice Weighted Shortest Job First (WSJF)

Scenario-based questions with a full explanation on every answer — including why the distractors are wrong.