Run a backlog one job at a time and every job bleeds its cost of delayWhat it costs you per week to not have this yet — lost revenue, carried risk, a promise going stale. The same number that feeds WSJF. for every week it sits unfinished. So the sequence sets the total bill — and the cheapest order isn't “biggest first”. Weighted Shortest Job First — cost of delay ÷ duration — is provablySmith's rule (1956): sequencing by the ratio of weight to processing time minimises the total weighted completion time. WSJF is this rule applied to a backlog. the least-cost sequence.
| Sequence | Total delay cost |
|---|---|
| As plannedthe order in your backlog | 400 |
| Biggest firsthighest cost of delay first | 363 |
| Shortest firstquickest jobs first | 283 |
| WSJFcost of delay ÷ durationcheapest | 277 |
Running this backlog WSJF-first costs 277 in delay, against 400 for the planned order — 123 saved, the same work in the same weeks, just resequenced.
Read left to right: WSJF front-loads the jobs that bleed the most cost of delay per week of work, so the costliest waiting ends soonest. “Biggest first” looks responsible but parks a long job up front where it blocks every quick win behind it — which is why it's usually the dearest order of the four.