Prioritisation under constraint

Prioritisation is the discipline of deciding what not to do. Demand always exceeds capacity, so the job is to sequence work so that the most important outcomes happen first, and to say no to the rest clearly and early. Ranking a list is the easy part. The competency is the habit of reasoning explicitly about value and cost of delay instead of reacting to whoever asked loudest or most recently, together with the nerve to show the resulting trade-off to the people who lose by it. Without it an organisation has no priorities at all: everything is started and nothing finished, and a team or a whole organisation stays fully busy while producing little. The examinations test it at every scope for two reasons. The failure is easy to recognise (the yes to everything, the thirty priority-ones, the failing initiative kept alive by its sponsor), and the defensible answer nearly always brings to the surface a trade-off that the comfortable answer quietly absorbs.

1. Sequence by value and cost of delay, not by advocacy. Ask what the organisation loses for each week a piece of work goes unshipped, compared with the alternatives. Who asked, how loudly and how recently are not inputs.
2. Say no by surfacing the cost of yes. A strong no is never a flat refusal. It takes the form of trade-off transparency ("we can do this, and that slips. Which do you want?"), so the stakeholder makes an informed choice and the manager has not silently absorbed the cost.
3. Starting less finishes more. Work in progress trades against throughput. The more that is started, the slower everything finishes, and a system loaded to full utilisation has no room for the surprise that is always on its way. Looking busy and being effective are different things.
4. Defend focus against churn. Reprioritising so often that nothing ships is a failure of prioritisation, and should not be mistaken for responsiveness. A priority that changes every week is not a priority.
5. Own the trade-off at the level where the conflict lives. When teams compete for a shared bottleneck, or resist lending people to an initiative, the manager above them decides against a single ranked view and names what slips. Refereeing the dispute is not the same as deciding it.
6. A portfolio is small enough to finish. At organisational scale, adding parallel initiatives does not add throughput. It adds the appearance of progress while everything slows. The intervention is fewer concurrent initiatives, explicitly ranked and sequenced so they finish.

The economics of prioritisation rest on queueing theory, which is mathematics. Its application to management rests on practitioner work with clear provenance, so the unit grades it as practice.

Cost of delay practice

Cost of delay is the value lost for each unit of time a piece of work stays unshipped. Donald G. Reinertsen, The Principles of Product Development Flow: Second Generation Lean Product Development (2009) sets it out for product development as the single number that should drive sequencing. Its strength is that it turns arguments about importance into arguments about economics: work with a high cost of delay and a short duration goes first, however senior its advocate. Most delay costs are estimates. The discipline is to treat them as explicit assumptions that people can argue about, and to avoid the false precision that dresses opinion up as arithmetic.

Work-in-progress limits and flow practice

The same source supplies the queueing result that a system loaded to full utilisation has unbounded queue times, and its practical corollary: constrain how much is in flight so the most important things finish sooner. Starting less finishes more. The limits of the model are operational. A work-in-progress limit has to be enforced at intake or it becomes decoration, and the slack it creates has to be defended as capacity instead of being harvested as if it were laziness.

Urgent versus important practice

This distinction was popularised in Stephen R. Covey, The 7 Habits of Highly Effective People (1989) and is attributed to a remark of Dwight Eisenhower's. It separates what demands attention now from what matters most, because urgency systematically crowds out importance unless someone deliberately resists it. It is a lens and does not schedule anything. Real work is often both urgent and important, and what the distinction asks of a manager is to fund the important work that makes no noise. Labelling quadrants is beside the point.

Trade-off transparency practice

This is a practice with no formal name, and the standard treats it as the core of saying no. State what the request costs in terms of what else moves, and let the person who wants it choose. The refusal becomes a decision the stakeholder owns, the manager stops silently absorbing demand, and the reasoning behind the queue becomes visible. Its provenance is ordinary practice. Its absence is the most common prioritisation failure the examinations present.

Portfolio governance practice

Portfolio governance is the organisational form of the same discipline. It means an explicit, ranked and deliberately small set of concurrent initiatives, reviewed on a cadence against evidence, with the ranking used to settle contention for resources by strategy instead of politics. It has no single author. Every organisation that finishes things has converged on it, and "thirty priority-ones" is what its absence looks like.

The ranked queue with reasons

The team holds its work in one ordered list. Beside each item near the top sits the reason it is there: the outcome it serves and what delaying it would cost. A new request gets placed in the list at a particular position, and that placement is the conversation with the requester. If nobody can explain the order of a queue, it has been accumulated rather than prioritised.

The cost-of-yes conversation

When a stakeholder asks for something new, the manager answers with what would move: "yes, and this slips to next month. Is that the trade you want?" The stakeholder decides. The manager records the decision where the next stakeholder will see it, so the history of the queue is visible and the trade-offs build up into a record instead of a grievance.

Limiting what is in flight

The team agrees how many things it will have open at once and holds to that number at intake. Once the limit is reached, something finishes before anything new starts. Enforcement matters more than the number itself. A limit that gives way whenever someone senior asks is only a suggestion.

One voice to a shared dependency

When several teams depend on the same bottleneck, their manager sets one ranked list across them and represents it once, which ends the internal competition. If each team escalates its own request separately, that is three escalations and no decision.

The portfolio review

At organisational scale this is a regular review, based on evidence, of the initiatives in flight: what each is meant to produce, whether it is producing it, and what the ranking implies when they contend for the same people. Sometimes the outcome is a stop. A clearly failing initiative gets the honest recommendation whoever sponsors it, because sponsorship is no reason to keep spending the organisation's capacity.

At team scope the object is the team's own queue. The manager sequences it directly and protects one team's focus against noise. At organisational scope the object becomes the competing priorities of several teams and the shared bottleneck they contend for. The characteristic move is a single ranked view, represented with one voice, with work in progress limited across the group and the staffing trade-offs owned by the manager instead of refereed. At executive scope the object is the portfolio: which bets the organisation makes, which it declines and which it stops, with the ranking settling contention for scarce senior people by strategy, whoever happens to be shouting. How Judgment Scales describes the pattern, and SV-1 Engineering strategy covers the strategy the portfolio serves.

• Holds an explicitly reasoned queue and can say why each item near the top is there. team
• Surfaces the cost of yes and lets the stakeholder choose, instead of absorbing the request. team
• Defends the team's focus against the churn of constant reprioritisation. team
• Recognises that a team with ten things in progress and nothing shipped has a prioritisation problem, not a capacity problem.
• Failure mode: loudest-voice prioritisation. team
• Failure mode: saying yes to everything and quietly missing everything. team
• Failure mode: treating every "urgent" as important. team
• Failure mode: reprioritising so often that nothing ships. team
• Sets one ranked view of demand across several teams and represents it to shared dependencies with one voice. org
• Limits work in progress across the group so that the most important things finish. org
• Owns the staffing trade-off for a cross-team initiative (decides against the department's priorities and names what slips) instead of refereeing three managers' reluctance. org
• Fixes intake as a system: one visible front door, stakeholders redirected consistently, engineers backed when they redirect. org
• Failure mode: letting each team fight its own escalation war. org
• Failure mode: measuring health by how busy every team is. org
• Failure mode: punishing engineers for accepting work the intake system failed to catch. org
• Keeps the portfolio small enough to finish, explicitly ranked and sequenced. exec
• Uses the ranking to settle contention for scarce senior people by strategy, once, instead of by sponsor. exec
• Recommends stopping or descoping a failing initiative regardless of its sponsorship. exec
• Leads a crisis reallocation through explicit priorities and through leaders, over-communicating the reasoning. exec
• Failure mode: thirty priority-ones. exec
• Failure mode: measuring the organisation by initiatives started. exec
• Failure mode: letting sponsorship immunise failing work. exec
• Failure mode: personally micromanaging a crisis shuffle instead of leading it. exec

Focus versus responsiveness. A queue that never changes ignores new information, and one that changes every week ships nothing. The judgment lies in how big a change has to be before it justifies re-sequencing, and in making each re-sequence an explicit decision with a stated reason instead of a reflex.

Local versus global. If every team maximises its own throughput, the whole falls out of alignment. If every decision is taken centrally, the parts slow down. The answer depends on where the ranking lives: priorities and boundaries are set at the level that can see the whole, and the teams sequence their own work inside them.

Utilisation versus flow. A fully busy team looks efficient and is one surprise away from missing everything. Deliberate slack looks like waste, yet it is what lets the important things finish on time. Part of the competency is defending that slack against people who read it as idleness.

Sponsorship versus evidence. A failing initiative with a powerful sponsor is politically expensive to stop and organisationally expensive to continue. The honest recommendation is owed whoever the sponsor is. The judgment is in how it is delivered: with the evidence, with options, and early.

Decisiveness versus consensus. In a crisis, a clean priority call made quickly beats a consensus reached late. In steady state, a ranking imposed without hearing the teams will be resisted and gamed. The scope of the situation decides which mode applies. The failure is running in crisis mode all the time, or seeking consensus in the middle of a crisis.

A head of engineering inherits an organisation of twelve teams with thirty initiatives in flight, each labelled a top priority by someone. Nothing has finished cleanly in two quarters. Every team is fully allocated, several engineers are on three initiatives at once, and two strategic programmes, whose sponsors both sit on the executive team, are fighting over the two most senior engineers in the organisation. The previous head of engineering responded by adding a programme office and a weekly status meeting, which produced a very good dashboard.

The temptation is to fix the coordination with better tracking, clearer dependencies and a more senior programme manager. That treats the symptom. Thirty concurrent initiatives across twelve teams is an absence of prioritisation, and no amount of visibility will make them finish. At this scale, adding parallel initiatives does not add throughput. It adds the appearance of progress while everything slows down.

The head of engineering forces a real ranking. Working with the executive team, they list the thirty initiatives, attach to each the outcome it serves and the cost of delaying it by a quarter, and rank them. Then they draw a line. The organisation will run the top eight to completion, sequence the next ten behind them, and stop or fold the rest. The stopped initiatives are named, with the reason, to their sponsors and to the organisation, because an initiative that is quietly starved does more harm than one that is explicitly stopped. The two senior engineers go to the higher-ranked of the two strategic programmes, and the other sponsor hears the reason from the ranking instead of from a negotiation.

Two further choices are deliberate. The ranking is published, so the next contention for a scarce person is settled by looking at the list without anyone escalating. The programme office stays, but its weekly meeting changes purpose: it stops reporting status and starts examining evidence against the ranking, asking whether the top eight are producing what they were meant to and whether any of them should stop.

The costs are real. Three sponsors are unhappy, one initiative with a powerful backer is among those stopped, and the organisation spends a month re-sequencing. What the head of engineering does not do is keep thirty things alive to avoid that unhappiness, because that choice is what produced two quarters in which nothing finished.

• DE-2 Estimation, planning, and honest commitments: the commitments that follow from the sequence, and the honesty about uncertainty they require.
• DE-4 Managing scope, risk, and dependencies across workstreams: the dependencies and shared bottlenecks a ranked view has to account for.
• SV-1 Engineering strategy: the strategy the queue and the portfolio exist to serve.
• SV-3 Roadmapping: roadmapping as the same trade-off between product delivery and platform investment, over a longer horizon.
• PL-1 Delegation, ownership, and decision rights: delegating the sequencing within boundaries the manager sets.

1. Why is "we can do this, and that slips. Which do you want?" stronger than a flat no?
   Answer

   It turns a refusal into a decision the stakeholder owns, makes the trade-off visible instead of absorbing it, and puts the reasoning behind the queue on the record. A flat no invites escalation, whereas a visible cost invites a choice.
2. A team has ten things in progress and has shipped nothing in three weeks. Is the problem capacity?
   Answer

   Almost never. Work in progress trades against throughput, and starting less finishes more. The intervention is a limit on what is in flight, enforced at intake. More people will not fix it.
3. Three of your teams each escalate a competing request to the same platform team. What does strong judgment do that forwarding the escalations does not? org
   Answer

   It sets a single prioritisation across the three teams and represents it to the platform team once, which ends the internal competition. Forwarding three escalations amounts to no decision at all.
4. Three managers each refuse to lend people to a cross-team initiative. Whose trade-off is it? org
   Answer

   Yours. Decide against the department's priorities, name what slips as a result, and own that cost. If you referee the reluctance, the initiative starves politely.
5. Why does adding parallel initiatives reduce an organisation's throughput, and what does that imply for portfolio size? exec
   Answer

   People and shared dependencies are then contended for across more work. As utilisation approaches full, queue times grow without bound, and everything slows while the appearance of progress rises. The portfolio should be small enough to finish: fewer concurrent initiatives, explicitly ranked and sequenced.
6. An initiative is clearly failing and has a powerful executive sponsor. What is owed? exec
   Answer

   The honest recommendation (stop or descope), with the evidence and options, early. Sponsorship is no reason to keep spending the organisation's capacity. Whether to deliver the recommendation is not in question; the judgment is in how it is delivered.
7. What is the difference between leading a crisis reallocation and micromanaging one? exec
   Answer

   Leading means making the priority call cleanly, delegating the execution of the shuffle to leaders, and over-communicating the reasoning. Micromanaging means personally moving people, which stops the executive from running the system that runs everything else.

• Donald G. Reinertsen, The Principles of Product Development Flow: Second Generation Lean Product Development (2009): cost of delay; work in progress, utilisation and queue times.
• Stephen R. Covey, The 7 Habits of Highly Effective People (1989): the urgent-versus-important distinction as popularly framed.