By the time a project is late, the portfolio decision that made it late is eighteen months old.

A delivery review is a comforting place to be. The problem has a name, an owner and a recovery plan. Someone is accountable. The conversation is concrete.

It is also, very often, the wrong conversation. When four projects in a portfolio slip in the same quarter, and they share a scarce specialist team, a single integration platform or one overcommitted executive sponsor, the cause is not four delivery failures. It is one portfolio decision — made at selection, when the organisation approved more work than it could execute and recorded that approval as ambition.

The distinction is not academic. Delivery interventions applied to portfolio problems reliably fail, because they attempt to recover throughput from a system that has no spare capacity to give. Adding governance to an overloaded portfolio makes it slower, not faster.

The Strategic Context

A portfolio is not a collection of approved projects. It is a claim on a finite set of shared resources over time — people, funding, systems, executive attention, organisational tolerance for change — and the central portfolio question is not "is this a good initiative" but "can this organisation absorb this initiative alongside everything else it has already promised".

Most selection processes answer the first question well and the second not at all. Business cases are assessed individually, in sequence, against criteria that are internal to the initiative: strategic alignment, benefit, cost, risk. Each may pass cleanly. The portfolio can still be undeliverable, because the property that makes it undeliverable — concentration — is invisible from inside any single case.

This is the structural weakness. Individual rigour does not aggregate into portfolio rigour. [Related article: Strategy Is a System of Choices, Not a Document]

What Leaders Commonly Misread

That portfolio management is prioritisation. Ranking is one input. Balance is the actual discipline: balance between horizons (run, improve, transform), between risk profiles, between parts of the business, and above all between demand and capacity. A perfectly ranked portfolio that exceeds capacity by forty per cent will still fail, and it will fail from the bottom of the list upward in an order nobody chose.

That capacity means budget. Funding is the easiest constraint to relax and the least likely to be binding. The binding constraints are usually a named group of twenty to fifty people — integration architects, senior estimators, regulatory specialists, the two engineers who understand the legacy control system — plus the change absorption capacity of the operational areas being asked to adopt something new. Neither appears on a funding sheet.

That dependencies are a delivery concern. A dependency between two projects is created at selection, not at delivery. Approving Project B on the assumption that Project A will have delivered a platform by Q3 is a portfolio decision that binds two initiatives together. Delivery teams inherit it and are then held responsible for it.

That a portfolio can be optimised once. Portfolios are not optimised; they are continuously rebalanced. An annual selection round with no in-year reallocation mechanism guarantees that by month seven the organisation is executing last year's judgement against this year's conditions.

Reframing the Issue

The useful reframe is to stop asking what should we fund and start asking what shape should this portfolio have.

Shape has four dimensions that a ranked list cannot express.

Horizon balance. How much of the portfolio sustains the current business, improves it, and changes what business we are in? Organisations under margin pressure drift toward run-and-improve, which is rational quarter to quarter and quietly fatal over a decade. The drift is invisible without deliberate measurement, because no individual decision causes it.

Risk concentration. Are the initiatives carrying the largest benefit also the ones that share the same technology, the same supplier, the same regulatory assumption or the same handful of people? Concentration is how a manageable risk becomes an existential one, and it is only visible in aggregate. [Related article: The Three Levels of Risk Executives Rarely See Together]

Capacity load. Against the genuinely scarce resources — not the average resource — is the portfolio over-committed, and by how much and when? This requires naming the scarce groups explicitly, which most organisations resist because it makes a small number of people visibly critical.

Time-to-value distribution. If every initiative returns value in year three, the portfolio has no self-funding mechanism and no early evidence that its assumptions hold. A portfolio needs some short-cycle work not because short-cycle work is better but because it generates information.

The Capacity Conversation Nobody Wants to Have

Portfolio capacity analysis fails for a social reason more than a technical one. Making capacity explicit means naming which individuals are the constraint, and that has three uncomfortable consequences: it identifies a retention risk, it exposes a succession gap, and it implies that some sponsor's approved initiative cannot start when they were told it would.

Organisations avoid this by using average capacity — total available hours against total estimated hours — which almost always shows the portfolio as feasible. It is feasible, in the sense that a hospital with enough beds in total is feasible while its intensive care unit is full.

The correction is narrow and specific. Identify the five to ten scarce capability groups. Model the portfolio's demand against those groups only, by quarter. Ignore everything else. The analysis is cruder and far more useful, because the answer it produces — "the integration team is committed at two hundred per cent from March" — is actionable in a way that a blended utilisation figure never is.

Expect the first run of this analysis to show an over-commitment between thirty and eighty per cent. That is normal and is not evidence of poor planning; it is evidence that the organisation has been selecting initiatives one at a time. [FACT CHECK REQUIRED — the typical over-commitment range should be verified against organisational data before publication as a general claim.]

Decision Framework

Apply these five tests to a portfolio as a whole, not to its components.

1. The scarce-resource test. For each of the five most constrained capability groups, is committed demand below available supply in every quarter of the next four? Where it is not, which initiatives are being deferred, and has that deferral been stated rather than left to emerge?

2. The concentration test. If the single largest shared dependency — one supplier, one platform, one regulatory outcome, one person — failed, what proportion of the portfolio's expected benefit is affected? Above roughly a third, the portfolio has a concentration problem regardless of how well each initiative is managed.

3. The horizon test. What proportion of investment sits in run, improve and transform? Is the split deliberate, and has it moved over the last three cycles in a direction anyone chose?

4. The information test. Which initiatives will produce decision-useful evidence within two quarters? If the answer is none, the portfolio cannot learn and every commitment is effectively irreversible until it completes.

5. The stopping test. What was stopped, paused or descoped in the last cycle? A portfolio in which nothing stops is not being managed; it is being accumulated. [Related article: Stopping Work Is a Capability, Not a Failure]

From Strategy to Execution

Immediately. Add one question to every investment approval: which named scarce group does this consume, and in which quarters? Approvals that cannot answer are deferred, not rejected. This single change surfaces most concentration problems within a cycle, at almost no cost.

Over one to two quarters. Build the scarce-capability model described above and run it against the existing approved portfolio. Present the result to the executive team as a portfolio decision, not as a delivery risk report — the required action is reallocation, which only the portfolio authority can take. Establish an in-year reallocation mechanism with a standing agenda slot, so that rebalancing is routine rather than exceptional.

Over one to three years. Address the concentration itself, which is a capability problem wearing a portfolio costume. Where three initiatives depend on two integration architects, the portfolio answer is sequencing; the enterprise answer is a third architect, a simplified integration pattern, or a decision to stop building things that need one. Only the enterprise answer changes the constraint. [Related article: The Capacity Ceiling: Why Funded Portfolios Still Stall]

Signals to Monitor

  • Slippage clustering in time rather than spreading. Several initiatives slipping in the same month points to a shared constraint, not to shared incompetence.
  • Rising proportion of initiatives in "started but not progressing". The classic signature of demand exceeding capacity: work is begun to show commitment, then starved.
  • Recovery plans that all request the same people. Confirms the constraint and shows that recovery is being planned at project level against a portfolio problem.
  • Benefit forecasts revised down without initiatives being stopped. The portfolio is quietly becoming unfundable while continuing to consume capacity.
  • Approval volume rising while completion volume is flat. Work-in-progress is accumulating and cycle time will lengthen.
  • The same sponsor named on more than three transformational initiatives. Executive attention is a scarce resource with no utilisation report.

Questions for the Leadership Team

  1. Which five capability groups are genuinely scarce in this organisation, and what is our committed demand against each over the next four quarters?
  2. If our largest shared dependency failed tomorrow, what proportion of forecast portfolio benefit would be at risk?
  3. What did we stop last cycle, and what did the released capacity actually go to — or did it silently absorb overruns?
  4. Is our run/improve/transform split the one we intended, and who would notice if it drifted?
  5. Which of our current initiatives will tell us something decision-useful within six months?
  6. When a project slips, what is our first question — about the project, or about the portfolio?
  7. Who holds the authority to reallocate funding and people mid-year, and when did they last use it?

Closing Perspective

Portfolio management is often described as choosing the right investments. That framing flatters the selection decision and obscures the harder one. The right investments, chosen one at a time without reference to each other, still produce a portfolio that the organisation cannot execute — and the resulting failures will be reported, investigated and remediated as delivery problems, which is where they are least tractable and most expensive.

The leadership responsibility is to hold the aggregate view that no project manager can hold and no business case can express: how much this organisation can actually absorb, where it is concentrated, and what has to be given up so that the rest can succeed. That view is uncomfortable to construct and unpopular to act on. It is also the only place where portfolio failure can still be prevented rather than explained.

Next in this series: [Related article: Governing a Transformation You Cannot Fully Specify] — what governance must do once a portfolio becomes a program of interdependent change.