Method note 02 · September 2026

The wrong foundation

One discipline, two foundations

Ask a portfolio management office what its portfolio returned last year and you will usually get a delivery statistic. Seventy percent on time. Eighty percent on budget. Both are answers to a question about execution. Neither is an answer to the question that was asked.

Ask a private equity firm the same thing and you get returns by vintage, by class, measured against a benchmark, with an account of what worked and what did not. The difference is not diligence. Both organizations are staffed by capable people doing what their instruments were built to do.

The difference is ancestry. Deciding which work to fund is a question about allocating scarce capital between competing uses — the founding question of economics. But the discipline that grew up to answer it did not come from economics. It came from the project management office, and it arrived carrying project management's tools: intake forms, scoring rubrics, stage gates, status reports. Those are excellent instruments for controlling delivery. They do not become instruments of allocation by being pointed at a longer list.

Two frames of mind: portfolio management as an extension of project management, and portfolio management as economics Panel one: a straight chain from requests to score and rank to approve to deliver to report. Strategy is attached to the scoring step as a label. The budget enters as a fixed limit. Nothing returns. Panel two: strategic drivers set direction for ideas, provide the input to valuation at the pipeline, and bind the solver at the portfolio. A constraint boundary creates the boundary. Two dashed feedback paths return to the drivers, carrying what the pipeline supplied and what the portfolio funded. ONE The inherited frame Strategy applied as a label Requests received Score and rank ordered Approve funded Deliver status Report Budget a fixed limit One direction. Strategy is applied, not binding. The budget is a wall. Nothing returns. TWO The economic frame Strategic drivers Ideas Pipeline opportunities Portfolio Constraint boundary proposed candidates input to valuation binds the solver creates the boundary committed sets direction what the pipeline actually supplied what the portfolio actually funded Strategy binds the work. The constraint is priced. What was decided returns to what was declared. forward feedback
Figure 1. The same territory, drawn twice.

Where the models diverge

Value has units, or it does not. In the inherited frame value is a score. An initiative scores 736; another scores 507. The numbers order a list and do nothing else — 736 is not 1.45 times anything, and it cannot be set against a dollar. In the economic frame value is a quantity that can be compared with cost, and the moment it can, the marginal question becomes available: is the next dollar worth more spent here than withheld? That question is the whole of capital allocation, and a score cannot answer it. A non-quantitative marginal value has no meaning; it cannot be weighed against a marginal cost.

Strategy binds the work, or labels it. In the inherited frame strategy is an attribute of a request — a dropdown, a tag, a column on an intake form. It is applied to the work and then plays no further part, and adherence to it, if measured at all, is audited afterwards. In the economic frame the declared strategy is a preference structure that enters the selection itself. It is not checked after the fact. It constrains what can be chosen.

The constraint is a price, or a wall. In the inherited frame the budget is a hard limit; work is approved until the money runs out, and the limit is a fact of the environment, like weather. The frame has no concept of a constraint binding — of a limit that is actively costing something at the moment it is respected. In economics a binding constraint has a dual, which is to say a price: relaxing it by one unit is worth a specific, computable amount. An organization that cannot state what its budget ceiling is costing is negotiating its own capital allocation blind.

The loop closes, or stays open. The inherited frame is linear by construction. Requests are received, scored, approved, delivered, reported. Everything moves in one direction, and status travels forward to executives while nothing travels back to the strategy that started it. A system with no measurement of its own effect cannot correct itself; it can only be executed and then described.

An economic system is closed-loop. Decisions have consequences that return as information, and the return is not a courtesy — it is what makes the next decision better than the last. Two things close the loop here, and both are computable the day a portfolio is committed: what the pipeline actually offered against each declared priority, and what the money actually went to. Neither requires waiting years for outcome data. The inherited frame does not omit them because they are hard. It omits them because nothing in its lineage suggested that anything should come back.

The mistake

Almost nobody chooses the wrong frame. It arrives with the job.

The mistake is believing the difference is philosophical — a matter of emphasis, or maturity, or how sophisticated one wishes to be. It is not. The frame determines which questions can be asked at all.

An organization working in the inherited frame is not making worse decisions with the same information. It is structurally unable to produce certain information. It cannot say what its budget cap costs, because it has no value unit to price the cap in. It cannot say which commitments are fragile, because it has no notion of a solution being perturbed. It cannot compare what it declared with what it funded, because nothing returns. These are not gaps a better report would fill. They are consequences of the foundation.

The costs compound quietly. Consider what happens when a model is told to satisfy its constraints. It will happily fill the portfolio with work that consumes abundant, obsolete capacity while avoiding the initiatives that need the scarce new capability the organization is desperate to build. Every limit is respected. Every box is ticked. The portfolio is mediocre, and nothing in the process notices, because conformity and optimality were never distinguished.

Then there is addition. A scoring model sums the contributions of whatever it selects, which means it will always prefer several modest initiatives to one substantial one: they add to more and cost less. Phil Jackson recounts Jerry Buss, then owner of the Los Angeles Lakers, arguing Kobe Bryant out of a trade demand by asking whether a diamond of four or five carats should be given up for four diamonds of one carat. It should not, and no executive would think otherwise, because the value of a portfolio is not the sum of its parts. That assumption is rarely stated and frequently false, and a model built on it will systematically buy breadth the business does not want.

And decisions arrive binary and certain when they are neither — selected or rejected, above the line or below it. Executives know this is not how it feels, which is why they override the output, which is then treated as a governance problem rather than as accurate skepticism.

A different foundation

None of this requires abandoning the delivery discipline. It requires ceasing to use it to allocate.

Establish a value unit, and be honest about what kind it is. Most benefits cannot be expressed in money, and pretending otherwise is worse than not trying. But a subjective value can still be ratio-scale, elicited rather than asserted, and applied consistently. What matters is not that it is monetary. What matters is that it is a quantity on a defined scale, so that ratios and margins mean something.

Price the constraints. Compute the dual on whatever binds and report what one more unit is worth. A constraint whose price is unknown is a decision nobody has taken responsibility for.

Let the declared strategy enter the selection rather than auditing adherence afterwards. Adherence measured after the fact is a report. Adherence enforced during the solve is a contract.

And close the two loops that can actually be closed — declared priority against what the pipeline supplied, and against what the portfolio funded. Both are available the day the portfolio is committed.

What becomes possible

The question what is this constraint costing us? acquires an answer, and the budget conversation stops being a negotiation between assertions.

The decision becomes defensible in a specific sense: for any initiative, the organization can state what would have had to be different for the outcome to change. That is not the same as being right. It is considerably rarer.

Governance attention moves to the initiatives where the decision is genuinely uncertain, rather than spreading evenly across a list in which most entries were never in question.

And the declared strategy is finally confronted with the funded portfolio, which is uncomfortable and is the point. Either the organization failed to fund what it said mattered, or it does not believe what it declared. Both are worth knowing before the next planning cycle rather than after it.

What this note does not claim

Project management thinking is not wrong. It is right for delivery and wrong for allocation, and the error is one of application rather than of quality. Nor does economics dissolve the problem: rationality is bounded, judgment is biased, most benefits genuinely resist quantification, and organizations are political. The economic frame does not remove those difficulties. It makes them visible and, in some cases, priceable — a smaller claim than it may appear, and still a large improvement on not seeing them.

Nor are the vendors serving this market unserious. The lineage explains their instruments without blaming anyone, and it also explains why two decades of better dashboards, better workflow and better scoring have changed so little. They are refinements of the wrong object.

Ignoring economics is not portfolio management.

Cite as: Gruia, M. (2026). The Wrong Foundation: One Discipline, Two Foundations. Frontnovo Method Notes.