Three funded initiatives each need about six weeks of work from the same two-person identity team, which also absorbs roughly 30% of its time in unplanned support. All three sponsors want to start now and the plan shows all three starting now. Which rule should decide the sequence?
Show the full answer Hide the answer
The deciding property
The identity team is the constraint, so the only denominator that matters is its time. Ranking by value, by effort, or by total initiative size all give the wrong order when three pieces of work compete for one scarce resource. The rule is cost of delay divided by the weeks of constraint time the item consumes, which is the refinement that most prioritisation exercises miss: they divide by total effort, where the other 90% of effort is not scarce and does not belong in the ratio.
Worked, with the arithmetic visible. Suppose the three items carry cost of delay of £40k, £120k and £30k per month and need 6, 6 and 2 weeks of identity time respectively. Per week of identity time that is £6.7k, £20k and £15k. The order is the £120k item, then the small £30k one, then the £40k — and the two-week item jumping ahead of the six-week one is exactly the result a value-only ranking never produces.
The honest completion date
18 weeks of demand against a team that has 70% of its time available is about 26 calendar weeks before any coordination overhead, and the third initiative finishes at the end of it regardless of the order chosen. Sequencing does not make the work smaller; it decides who waits and it makes the waiting visible. Saying so plainly is most of the value of the exercise, because the alternative is three sponsors discovering in month four that their plans were arithmetic fiction.
Why the other options fail
- Start all three and split the team's time. The comfortable answer, and it makes every initiative late instead of two. Splitting attention adds context-switching loss and triples work in progress, so nothing finishes early enough to start collecting its benefit; the £120k item loses months of cost of delay to keep two smaller ones warm.
- Earliest committed launch date. Right when a date is genuinely immovable — a statutory deadline or a contracted go-live, which is a constraint rather than a priority. Wrong as a general rule, because it makes commitment a currency: whoever announces a date first gets the team, and the organisation learns to announce dates early.
- Split the team so each initiative has one engineer. Removes review, on-call cover and pairing from a domain where a mistake is a security incident. A two-person team split in two is not two teams; it is two single points of failure with no one to check their work.
What would flip the decision
| If this changes | Choose | Because |
|---|---|---|
| One item has a statutory date | Date order | The date is a constraint; ranking below it is optional work |
| The identity work is nearly identical across the three | Build the shared capability first | One 8-week investment removes the constraint from the other two |
| Cost of delay is unknown for all three | Sequence by smallest first | Shortest queue time overall while you learn the numbers |
| The constraint team is idle half the time | Stop sequencing | There is no constraint to optimise; the plan's problem is elsewhere |
When not to score at all
With two items, skip the scoring and just sequence them. Scoring frameworks earn their overhead above roughly ten competing items, where intuition stops tracking and the argument becomes about seniority. Below that, the arithmetic is a formality and the meeting is the cost.
And when every ranking puts the same team at the front of every quarter, prioritisation is the wrong tool: the answer is to grow, split or automate the constraint, because no sequence of a two-person queue serves a twelve-team organisation.