WSJF example: score five jobs with cost of delay
A step-by-step WSJF example for SAFe and programme work: score five jobs with cost of delay, divide by size, and see where the method breaks.

Here is a wsjf example you can copy. It is for product managers and product owners who have to put a backlog in order in a SAFe or programme setting. At the end you will have five jobs scored on cost of delay, divided by size and ranked, and you will know when to distrust the result.
In brief
- WSJF = cost of delay divided by job size.
- Cost of delay = value + time criticality + risk reduction (or opportunity enablement).
- Score on one scale: 1, 2, 3, 5, 8, 13, 20.
- Score one column at a time across all jobs, not one job at a time.
- Do the highest WSJF first, then check the order before you trust it.

Write down the formula
WSJF means weighted shortest job first. The idea is simple: do first the job that loses the most value for every unit of time it takes to finish.
Cost of delay is what waiting costs you. It has three parts. Value is what the job is worth to users or the business. Time criticality asks whether that value shrinks if you wait, for example because of a fixed date or a window that closes. Risk reduction or opportunity enablement asks whether the job removes a threat or opens options for later work. Add the three. Then divide by job size, the relative effort to get the job done.
List the jobs and fix one scale
Put the jobs side by side. They should sit at the same level: five features, or five epics, not a mix of both. WSJF compares like with like.
Use the same scale for every column, including size: 1, 2, 3, 5, 8, 13, 20. The gaps grow on purpose. Nobody can honestly tell a 13 from a 14, so the scale does not offer the choice.
Score one column at a time
Take the value column first. Look at all five jobs together. Find the one with the least value and give it a low score. Then place each of the others against it and against each other. When the column is done, move to time criticality, then risk reduction.
This is why scoring by column keeps the scale honest. When you score job by job, you form a gut feeling about the whole job and then spread three numbers over it. The columns stop being separate questions. The scale also drifts: the first job is compared with nothing, and the last is compared with whatever mood you were in by then. In a column, every score is made against the same other jobs in the same sitting. In a group, you also argue about one question at a time, which is much easier than arguing about a whole job.
Score job size the same way
Size is a column too. Score it relative to the other jobs, and ask the people who will do the work, not the people who want the result. Size covers the whole job, including testing and release, not just the build.
Add, divide and sort
For each job, add the three scores to get cost of delay. Divide by size. Sort from highest to lowest. The top of the list is what you start first.
Worked example
Five jobs, scored by column
The jobs and numbers below are made up to show the method. Each column was scored across all five jobs before moving to the next.
| Job | Value | Time | Risk | Cost of delay | Size | WSJF |
|---|---|---|---|---|---|---|
| B | 5 | 13 | 3 | 21 | 3 | 7.00 |
| A | 8 | 3 | 2 | 13 | 5 | 2.60 |
| D | 3 | 2 | 13 | 18 | 8 | 2.25 |
| E | 13 | 8 | 1 | 22 | 13 | 1.69 |
| C | 20 | 5 | 8 | 33 | 20 | 1.65 |
The order is B, A, D, E, C.
Job B is not the most valuable. It wins because the cost of waiting is high (time criticality 13) and it is small. Job C has the largest cost of delay by far, 33, and still comes last, because it is also the largest job. Job D is worth little to users but removes a lot of risk, so it climbs above E and C. That is the method doing its work: it shows that a big prize is not always the best next step.
Look at E and C. They score 1.69 and 1.65, which is a tie for any practical purpose. Treat them as a tie and decide on something else, such as dependencies or what the team can start now.

Check the sums yourself
To try your own numbers, or to see how the order moves when one score changes, use PikoPM's free WSJF calculator. Work through the example above first and check that you get the same order.
Where WSJF breaks
One item scored alone. A score only means something against other scores. If you score one request by itself, a 13 for value just means it felt big. There is nothing to anchor it, and a request that someone senior cares about will usually get high marks. Always score in a batch, and put the new item in the same columns as the work already waiting.
Job sizes nobody believes. Size is the divisor, so it moves the result more than any other number. Go back to the example. If the team decided that job C is a 13 and not a 20, its WSJF becomes 33 divided by 13, about 2.54. It would jump from last place to second. If a change of one step in size reorders your list, the order is not stable. Ask the people doing the work to size the jobs, and say out loud which sizes they do not believe. A guessed size that nobody owns makes the whole ranking look more exact than it is.
Other limits. WSJF does not see dependencies, team capacity or a hard date outside the time criticality column. It gives you an order to discuss, not a decision.
Before you publish the ranking
- All jobs are at the same level of detail.
- Every column was scored across all jobs before the next column began.
- Everyone used the same scale: 1, 2, 3, 5, 8, 13, 20.
- The people who will do the work gave the sizes.
- Near ties are marked as ties.
- You changed each size by one step to see whether the order holds.
- Dependencies and capacity are checked outside the score.



