Chapter 6: Resources and Dependencies
The project that was always on the critical path
Sofia runs a small data team. A transformation program asks her for two analysts for six months. The program director says the work is “on the critical path” and has the chief executive’s attention. Sofia offers one analyst for eight weeks while the team finishes a regulatory deadline. The director replies that anything less than two people will put the entire program at risk.
At the portfolio meeting, Sofia learns that three other teams received the same message. The program plan lists all of their contributions as critical, although several tasks do not begin for four months. Meanwhile, the program has kept a contractor budget in reserve and has not filled two approved roles. The director may be protecting a difficult schedule, but the language of dependency has made every resource discussion sound like a choice between full support and organizational failure.
Resources turn plans into work. They include money, people, time, equipment, data, permissions, physical space, attention, and the capacity to absorb change. Because resources are finite, allocation always creates winners, delays, and trade-offs. This makes resource decisions a natural site of influence.
Dependency creates leverage. A team that controls a central workflow, scarce expertise, or an uncertain problem can acquire influence beyond its formal rank. Classic organizational theory linked subunit power with factors such as centrality, ability to cope with uncertainty, and difficulty of substitution.1 These features are not abuses; they explain why the database administrator, regulatory specialist, scheduler, or local coordinator may be more consequential than an organization chart suggests.
The practical task is to distinguish real constraint from unreviewable assertion—and to design dependencies that serve the work without making the organization captive to one claim or holder.
A resource claim has four parts
Before examining the five patterns, break a request into:
- Need: what outcome or obligation requires a resource?
- Amount and timing: what capacity is needed, when, and for how long?
- Dependency: what becomes delayed, degraded, unsafe, or impossible without it?
- Alternatives: what can be changed, substituted, staged, stopped, or accepted as risk?
“We need two analysts” is a request. “Without two analysts from October through December, migration testing finishes three weeks after the regulatory date; one contractor or removal of the reporting work would close the gap” is a resource case. It exposes assumptions and choices.
1. Overstating a dependency
A dependency exists when one piece of work relies on another. Plans need dependencies because sequence matters. A dependency is inflated when a task, person, system, or approval is described as more central or exclusive than the evidence supports.
Inflation can be deliberate, but it often arises from defensive planning. Program leaders fear that shared resources will disappear, so they mark everything critical. Specialists know that early warnings are ignored, so they describe a constraint in absolute language. Teams include generous buffers because the organization routinely changes priorities. Each actor behaves rationally within a low-trust system, and the portfolio becomes impossible to interpret.
Test a dependency with counterfactuals:
- What exact task stops if the resource is unavailable?
- For how long?
- Can the work be resequenced?
- Can scope, quality, or timing change?
- Is another qualified provider available?
- What risk would the decision owner accept?
- What evidence shows the requested amount?
Do not demand false precision. Novel work contains uncertainty. Use ranges and scenarios: one analyst produces a basic migration by December; two enable full reconciliation; zero delays compliance testing. The choice becomes visible without pretending the forecast is certain.
If every task is labeled critical, require the program to rank them. “Which three dependencies would you protect first if only half the capacity existed?” forces differentiation. A real critical path is a sequence, not a synonym for importance.
2. Tying an unrelated request to an accepted priority
Priority bundling occurs when a resource request borrows the legitimacy of an accepted goal without a sufficiently close connection. A discretionary dashboard is called a regulatory requirement. Additional headcount is attached to a customer-safety program although most of the role serves ordinary growth. A preferred office renovation is presented as necessary for return-to-work compliance.
Real initiatives have shared costs. A platform built for one requirement may sensibly serve others. The issue is attribution. What portion of the request is necessary for the accepted priority, what portion creates additional value, and who is authorized to choose the extra investment?
Use a component case:
| Component | Required outcome | Basis | Incremental benefit | Decision owner |
|---|---|---|---|---|
| Core access controls | regulatory compliance | cited standard and assessment | lower breach risk | risk committee |
| Analytics module | operational insight | optional design | faster reporting | portfolio board |
The table does not oppose the analytics module. It prevents the urgency and authority of compliance from deciding an ordinary investment by association.
Individuals should be careful when leaders say a request is “for” a high-status priority. Ask which requirement, deliverable, or accepted objective it satisfies. A clear sponsor will usually welcome the chance to show the link. If the connection is partial, separate the necessary minimum from the wider proposal.
3. Claiming scarcity without a reviewable basis
Scarcity is real. Budgets close, specialists are limited, equipment has finite capacity, and people cannot sustain unlimited workload. Yet scarcity claims often arrive as facts with no visible unit: “There is no budget,” “The team has no bandwidth,” “Only one slot exists.”
A reviewable claim states the relevant pool, period, commitments, and authority. There may be no discretionary budget in the current quarter, while emergency funds exist under another approval. A team may have no unallocated hours, while the organization can stop lower-priority work. A training class may have one place under the current booking, while another session could be purchased.
The alternatives may be unattractive. That does not make them nonexistent.
Do not respond by assuming hidden abundance. Capacity data is imperfect, and publishing every budget line can breach confidentiality or overwhelm. Ask for enough evidence to support the choice: the applicable limit, known commitments, priority rule, and next review point.
Managers should distinguish unavailable from not prioritized. “We do not have capacity” often means “We have committed capacity elsewhere and are not changing those commitments.” The second statement owns the allocation. It also lets others challenge the priority rather than arguing about whether humans possess more hours.
Scarcity should include workload health. Capacity is not the theoretical maximum before collapse. Sustainable planning accounts for routine work, coordination, leave, learning, error recovery, and unexpected demand. The World Health Organization’s guidance on mental health at work emphasizes organizational interventions and psychosocial risks rather than treating strain only as an individual resilience problem.2
4. Reserving resources before the stated process
Precommitment appears when money, staff, space, or capacity is reserved for one option before the organization completes its announced allocation process. The reservation can make a later review symbolic: other proposals compete for what remains.
Early reservation is sometimes prudent. Long-lead equipment, specialist availability, and expiring grants may require provisional holds. A contingency reserve protects emergencies. A team may begin low-cost preparation while approval is pending. The status and reversibility of the commitment matter.
Ask:
- Is the resource held, contracted, spent, or merely forecast?
- Who authorized the reservation?
- What event releases or confirms it?
- What cost arises if the final decision differs?
- Are other applicants told which pool remains?
A provisional reservation should have an expiry and a clear label. “Subject to portfolio approval” must mean that rejection remains practically possible. If cancellation would trigger a large fee or public embarrassment, the organization has already made part of the decision and should record it.
Budget owners can publish allocations in stages: mandatory commitments, contingency, prior authorized work, and genuinely available capacity. This reduces the recurring surprise in which a committee is invited to prioritize a total budget that is mostly spoken for.
5. Holding back available capacity
Capacity withholding occurs when a person or unit has usable resources within its authority but does not release them, without a reason the process can examine. It may involve staff, information access, equipment, funding, or approvals.
The phrase available capacity needs care. A team that appears idle may be on call, preserving resilience, doing invisible maintenance, or constrained by skills. A budget balance may be restricted. An employee may have hours but not the competence or authorization. Confirm the resource state before judging the choice.
There are also legitimate reasons to hold capacity: contingency planning, safety margin, uncertain demand, staff development, or protection from chronic overcommitment. The reason should be connected to an owner and rule. “We preserve 15 percent for incidents because service levels require response within two hours” is a policy. “We might need it” can become an indefinite veto.
Withholding can create bargaining power. A central team may respond quickly to favored work and slowly to routine requests, making relationships more important than priority. A manager may keep a vacancy open so it cannot be reallocated. A department may delay approval until another group adopts its preferred design.
The process remedy is a service and allocation model: request criteria, priority levels, response times, reserve policy, escalation, and capacity reporting at an appropriate level. This does not remove judgment. It turns unexplained discretion into bounded discretion.
Bottlenecks are organizational choices
Some dependencies arise from the work. Others arise from design. If one person is the only authorized approver, one team owns every integration, or one system holds all records, the organization has concentrated power and failure risk.
Sometimes concentration is efficient or required. Separation would duplicate rare expertise, weaken security, or create inconsistent standards. The question is how the organization contains the dependency. Backups, documented decision boundaries, service expectations, succession, and periodic review can preserve expertise without creating captivity.
Do not punish the bottleneck holder for being needed. They may be absorbing the consequences of years of underinvestment. “Why are you blocking us?” is less useful than “What demand enters your queue, what capacity exists, and who decides priority?” The answer may show that the supposed gatekeeper has no authority to reject work and no resources to complete it.
The organization should also examine incentives. If a specialist becomes less secure after sharing knowledge, documentation will always lose to urgent delivery. If a manager loses headcount whenever a team creates efficiency, apparent scarcity becomes rational self-protection. Resource transparency must be paired with fair treatment.
Sunk costs and continued investment
Resource politics does not end at initial allocation. Once an organization has spent money, reputation, and leadership attention, it may continue funding a struggling course because stopping would make the earlier choice look wasteful.
Experimental and organizational research has examined escalation of commitment: persistence in a chosen course despite negative feedback. A meta-analytic review found that multiple project, psychological, social, and structural factors are involved; no single explanation fits every case.3 This is important because casual accusations of “sunk-cost thinking” can be as simplistic as blind continuation. New investment may be rational if future benefits still exceed future costs.
Separate past and future:
- What value remains recoverable?
- What new resource is requested?
- What evidence has changed?
- What alternatives exist now?
- Who can evaluate without defending the original choice?
- What stop, pivot, or continuation criteria were set?
Do not ask whether the organization can “waste” what it already spent. That cost is largely gone. Ask what the next unit of time or money is expected to achieve.
Independent review can help when original sponsors are strongly identified with the project. Independence does not mean excluding their knowledge. It means that continuation is not decided solely by those whose standing is tied to the original decision.
Invisible resources and invisible work
Resource plans often count named roles and budgets while overlooking coordination, emotional labor, onboarding, translation, accessibility work, maintenance, and relationship repair. The omission creates two political effects. Some teams look more productive because another group carries their supporting work. Some employees appear to have capacity because their invisible contributions are not in the plan.
Ask what work makes the visible deliverable possible. Who schedules, checks, explains, cleans data, calms customers, trains newcomers, or repairs errors? Which tasks recur, and which are voluntary? Does the organization reward the work, rotate it, or allow people to decline?
Not every helpful act needs a time sheet. A quarterly workload conversation can surface patterns. List major outputs, recurring support, peak periods, and work that only one person can do. Compare demand with sustainable capacity. If the same employee always takes notes or mentors new staff, make the assignment visible and decide whether it belongs in the role.
This is also an equity issue. Research and practice have raised concerns that some groups carry disproportionate “office housework” or diversity labor, but a specific workplace conclusion requires its own evidence. Use local task data and close comparisons rather than assuming the pattern from demographic identity alone.
A response ladder for resource pressure
Translate the claim. Convert “critical,” “no bandwidth,” or “must have” into need, amount, timing, consequence, and alternative.
Request scenarios. Ask what happens with full, partial, delayed, substituted, or zero allocation.
Expose the priority choice. Identify what existing work would move, stop, or carry more risk.
Separate components. Distinguish mandatory minimums, optional value, contingency, and prior commitment.
Clarify authority. Determine who owns the resource, who sets priority, and who may accept residual risk.
Set a review point. Time-limit reservations, exceptions, emergency capacity, and continued investment.
Escalate health or safety promptly. If workload or resource restriction creates immediate danger or protected concerns, use the appropriate formal route rather than treating it as a routine prioritization debate.
For managers: make scarcity comparable
A portfolio cannot compare requests written in different currencies. One team asks for people, another describes revenue, a third invokes regulatory risk, and a fourth says a senior leader promised support. Create a common resource case:
| Field | What to state |
|---|---|
| Outcome | the result or obligation served |
| Capacity | people, money, time, equipment, or permission needed |
| Timing | start, duration, and deadline basis |
| Dependency | what fails or changes without it |
| Options | full, minimum, delayed, substituted, or stopped |
| Existing commitment | resources already reserved or spent |
| Risk owner | who accepts the consequence of underfunding or delay |
| Review | when allocation will be checked or released |
Then publish the decision at a level appropriate to confidentiality: which requests were funded, deferred, reduced, or declined and on what priority basis. Transparency does not require exposing salaries, personal circumstances, or sensitive strategy.
Sofia asks the transformation program to show its dependency scenarios. The analysis reveals that one analyst is needed immediately, a second in four months, and contractor funds can cover reconciliation. The portfolio board allocates Sofia’s analyst for eight weeks, requires the program to fill its approved roles, and sets a review before the second phase. The transformation remains important. It no longer has to pretend that every request is equally urgent.
The vocabulary of budget conversations
Budget language can hide where choice remains. A cost may be called fixed because a contract exists, although the contract has a break clause. A budget may be ring-fenced by policy, donor condition, law, or merely an internal plan. A role may be approved without funding to recruit it. Money may be spent, committed, accrued, forecast, or simply reserved. These states are not interchangeable.
General readers do not need to become accountants to ask useful questions. Ask what the term means in this process and what action has occurred. “When you say the funds are committed, has a contract been signed, or have we assigned them in the forecast?” The answer identifies reversibility.
Likewise, “under budget” does not automatically mean available money. A project may have delayed invoices or costs that cannot transfer between categories. “Over budget” does not automatically mean poor control; exchange rates, emergency work, or an approved scope change may explain it. The process should connect variance to cause and authority.
Leaders can reduce political ambiguity by publishing a small glossary and reporting budget status consistently. The aim is not financial theater. It is to ensure that a familiar word does not carry a different constraint depending on who is asking.
Negotiating shared capacity
Shared teams—legal, data, design, procurement, communications, security, human resources—often become the place where organizational priorities collide. A first-come queue favors early insiders. A loudest-voice system favors powerful sponsors. A purely numerical score may conceal judgment behind weights.
A workable service model separates classes of demand. For example: emergencies with defined criteria; mandatory or regulatory work; committed delivery; and discretionary improvement. Each class has an owner and service expectation. Within a class, the shared team can compare consequence, deadline basis, effort, and opportunity cost.
Requesters have duties too. They should involve shared teams early, provide a decision-ready request, respond to questions, and release capacity they no longer need. Reserving a specialist “just in case” blocks other work. Shared teams should be able to expire inactive requests after notice.
When priorities conflict, bring the conflict to the authority that owns the combined outcome. Do not require a security analyst to decide whether a product launch matters more than a finance migration if that trade-off belongs to an executive portfolio owner. The specialist should estimate effort and risk; the accountable leader chooses between business outcomes.
A monthly demand view can show the top commitments, available capacity ranges, queue age, and decisions needed. Avoid individual productivity rankings. The purpose is to allocate work, not monitor every minute.
Reserves, slack, and resilience
Unused capacity can look wasteful. In systems exposed to incidents, absences, demand spikes, or uncertain work, it can be the source of resilience. A hospital bed, cash reserve, backup supplier, unfilled calendar block, or spare server capacity may exist precisely because average demand is not the only demand that matters.
The challenge is to distinguish designed reserve from unexamined withholding. A reserve needs a purpose, size or range, activation authority, and review. For example: “The service team keeps one person-day per week unallocated for severity-one incidents; if unused by Thursday, it may support the improvement queue.” That rule protects response while preventing capacity from disappearing indefinitely.
Reserves can also become political shields. A leader may call every surplus contingency, while another team absorbs chronic shortage. Portfolio owners should review reserve levels across units, taking different risks into account. Equal percentages may be inappropriate; a transparent rationale is more useful.
During cuts, organizations often remove all slack and later celebrate employees for extraordinary recovery. Repeated heroics are evidence that the base design may be too brittle. Track not only whether deadlines were met but what overtime, deferred maintenance, skipped learning, or personal strain made them possible.
Scarcity and fairness during contraction
When resources genuinely shrink, procedural quality matters more, not less. People may lose projects, opportunities, hours, or jobs despite competent management. A fair process cannot remove the loss. It can make priorities, criteria, voice, explanation, and support more credible.
Avoid announcing that “everyone must share the pain” before examining unequal starting positions and consequences. A uniform percentage cut may leave one unit functional and another unsafe. A policy that eliminates travel may affect a remote field team differently from headquarters. Consistency is a consideration, not a substitute for impact analysis.
Leaders should state the constraint, decision authority, protected obligations, criteria, timeline, consultation purpose, and available support. If certain information cannot be shared, explain the category and decision consequence. Do not invite ideas after the main commitments are irreversible unless the remaining purpose is honest.
Employees should be allowed to identify hidden dependencies and impacts without being treated as resistant. They may know that a “nonessential” role performs a mandatory control or that removing administrative capacity transfers work to higher-paid specialists. Listen before finalizing the arithmetic.
Where employment rights, collective consultation, discrimination, accommodation, health, safety, or contractual obligations may apply, use the appropriate professional and representative processes. A resource framework cannot determine legal compliance.
A manager’s conversation with an overloaded employee
Resource allocation happens inside individual workloads as well as portfolios. When an employee says they have no capacity, do not begin by demanding proof of effort. Begin with the work.
List current commitments, deadlines, recurring duties, and known interruptions. Ask which tasks only they can do and which have hidden quality or safety requirements. Then make the priority decision together:
Of these six commitments, the client correction and payroll control come first. Pause the dashboard redesign. Move the training deck to me. Tell the project sponsor that the analysis will be Friday, not Wednesday, and copy me.
This response does more than express sympathy. It changes demand and owns the consequences. Advice to “manage time better” without removing or moving work leaves the scarcity untouched.
If overload persists, examine staffing, role design, process waste, skill gaps, and health support. Do not use an employee’s willingness to cope as evidence that capacity exists. Nor should a manager promise resources they cannot authorize. State what you can change, what you will escalate, and when you will return with an answer.
Practice: challenge one scarcity sentence
Write down a resource sentence you recently heard: “There is no budget,” “We cannot spare anyone,” “This is critical,” or “The capacity is already committed.” Rewrite it with blanks:
For [period and resource pool], [amount] is committed to [named priorities]. Without [requested amount], [specific consequence] occurs by [date]. We could instead [alternative], which would create [trade-off]. [role] is authorized to choose, and we will review on [date].
Fill what you know. Mark the rest as questions. The exercise turns scarcity from an atmosphere into a decision.
Resources determine which commitments survive contact with reality. In the second half of the book, we follow what happens when work is completed—beginning with who receives the credit and visibility it creates.