The Recovery Portfolio Cannot Be Rebuilt with Pre-Crisis Scores
A portfolio is not rational because every project has a score; it is rational when the organisation can explain what it will not start, what it will stop, and what evidence would change its mind.
The Pipeline Looked Healthy
The investment committee entered the room with forty-three proposals, each reduced to a score on a hundred-point scale. The top eighteen fitted the capital envelope. Their business cases showed positive returns, their sponsors had accepted the benefits, and their status sheets were ready for approval.
Six months later, eleven had started. Four were waiting for the same two business analysts. Three depended on a systems architect already committed elsewhere. Two had quietly changed scope because customer demand had not recovered as forecast. None had been stopped, because no individual project had deteriorated enough to lose its original approval.
This is how a portfolio can be numerically disciplined and economically irrational.
As organisations rebuild their investment pipelines after the credit crisis, the instinct is understandable: refresh the business cases, tighten the hurdle rate, rescore every proposal, and fund the strongest. That method offers order after disruption. It also assumes that the principal task is to select good projects from a stable field.
It is not. The principal task is to make a sequence of commitments under conditions that remain uncertain. Demand is returning unevenly. Credit is more expensive and less forgiving. Operating budgets carry little slack. The people required to deliver change are often scarcer than the capital authorised to pay for it. A ranking cannot settle those questions because they concern timing, dependency and reversibility, not merely merit.
The recovery portfolio should therefore be rebuilt around constrained capacity and staged commitment, with scoring restored to its proper place as evidence rather than verdict.
Why the Old Scores No Longer Decide the Right Question
Most portfolio scoring models were designed to compare proposals. They reward strategic alignment, expected return, risk reduction and regulatory necessity. Used with care, they expose weak cases and impose a common language on competing sponsors.
Their weakness is not arithmetic. It is the proposition hidden beneath the arithmetic: that each project can be valued largely on its own, then placed in a queue from best to worst.
Three mechanisms make that proposition unreliable in the present recovery.
- The estimates are more precise than the environment. A projected return of 18 per cent appears superior to 15 per cent, but the difference is meaningless if volume, price or funding assumptions can move by twice that amount. The score converts uncertainty into a single number and then conceals how fragile that number is.
- Rank does not establish sequence. The fourth-ranked proposal may require a data clean-up funded by the twelfth. The highest-ranked customer initiative may consume the same process specialists needed for a mandatory control change. A list can tell us which proposals look attractive; it cannot tell us which combination can actually move.
- Project boundaries hide portfolio economics. Sponsors calculate benefits inside their own cases, while duplicated infrastructure, repeated process design and competition for scarce staff sit between cases. The portfolio pays those costs even when no project owns them.
A recovery portfolio is not a larger version of a project approval meeting. It is a set of choices about scarce organisational attention, made in an order that preserves the ability to change course.
That distinction matters because uncertainty is not evenly distributed. A replacement of unsupported equipment may have a clear cost and an unavoidable date. A new channel may have a plausible strategic logic but a wide range of demand outcomes. Treating both through one weighted score creates an appearance of comparability at the expense of judgement.
The Constraint the Capital Budget Does Not Show
The most revealing portfolio conversations begin when the capital total is temporarily removed from the page.
Consider a composite investment pipeline assembled in late 2010. Thirty-eight proposals sought £31.8 million. The provisional envelope was £22 million, and the scoring exercise selected twenty-four proposals worth £21.6 million. On paper, the portfolio fitted.
A capacity ledger told a different story:
| Constraint | Available for change | Demand in selected portfolio | Consequence |
|---|---|---|---|
| Business analysis | 14 full-time equivalents | 23 full-time equivalents | Nine proposals could not be shaped or mobilised as planned |
| Systems architecture | 5 specialists | 11 simultaneous assignments | Design decisions became a queue hidden inside project schedules |
| Operations process leads | 7 experienced managers | 16 part-time commitments | Benefits design was delegated to staff without authority |
| Testing environments | 3 usable environments | 8 overlapping windows | Release dates competed even where project plans appeared independent |
The issue was not a £400,000 overspend against the envelope. It was that the approved portfolio required roughly one and a half organisations to deliver it.
The response was not to improve the decimal places in the scores. The proposals were recut into four explicit groups:
- eight unavoidable commitments, including control remediation and replacement of obsolescent equipment;
- six recovery investments with evidence of near-term demand;
- four capability options whose first stage could test an assumption without committing the full case; and
- twenty proposals to stop, defer or combine.
The resulting first commitment was £14.2 million, not £21.6 million. That reduction was not a retreat from growth. It released the analysts, architects and operating managers needed to make the chosen investments move. Two of the capability options received only enough funding to complete process design, supplier evaluation and a limited market test. Their later expenditure remained conditional.
By the second review, one demand assumption had strengthened and one had weakened. Funding moved accordingly without requiring anyone to pretend that the original annual plan had been correct.
The pattern I have observed is that portfolios recover their momentum when leaders make scarcity visible in units that operations recognise: named roles, decision hours, environments, release windows and management attention. Capital remains essential, but it is rarely the only binding constraint.
The Strongest Case for Scoring
There is a serious argument for retaining a single, comprehensive ranking. After a period of cancelled budgets and hurried cost reduction, sponsors need to know that investment decisions are fair. Finance needs a traceable basis for allocating capital. A common score disciplines advocacy, limits the influence of the loudest executive and gives the board a defensible record of why one proposal preceded another.
Those are substantial advantages. Abandoning common criteria would invite bargaining disguised as strategy.
But the answer is not to let the score govern more than it can know. A score is strongest when it tests the quality of a proposition: Is the strategic connection real? Are the benefits material? Which assumptions drive the return? Is the downside understood? It is weakest when it is asked to choose between unlike obligations, resolve dependencies, or determine how much uncertainty the organisation can absorb at once.
The mature alternative is structured judgement with an audit trail. Keep the criteria. Keep the business cases. Record the assumptions and the reasons for each decision. Then require the governing group to state explicitly where it has departed from rank because of capacity, dependency, timing or the value of learning.
That is not a relaxation of governance. It is governance taking responsibility for the decision instead of transferring responsibility to a formula.
“The spreadsheet can discipline the conversation, but it cannot carry the judgement.”
Rebuild the Portfolio as a Sequence of Commitments
A rationalised pipeline needs fewer approvals and more meaningful decision points. Three changes alter its character without requiring an elaborate new apparatus.
Begin with the Non-Negotiable Boundaries
Before ranking discretionary proposals, establish the demands that genuinely cannot be avoided: legal or control obligations already in force, essential maintenance, contractual commitments and the minimum investment required to keep critical operations reliable. These should not receive artificially high strategic scores merely to force them above the line. They should be identified honestly as constraints on choice.
Then state the scarce delivery capacities in the same meeting. If the organisation has five systems architects available for change, the portfolio is not entitled to assume eleven. If operating management can absorb only two major process changes in a half-year, approving five does not create capacity; it merely creates delay and divided attention.
This boundary-setting turns the argument from “Which projects deserve approval?” to “What combination can we responsibly undertake?”
Commit in Stages Where the Assumption Is Uncertain
The annual business case encourages a false choice between full approval and rejection. Many recovery investments are better treated as a sequence.
- Fund the smallest credible stage that can produce evidence: a design, a supplier response, a prototype, a customer trial or a validated cost baseline.
- State the assumption the stage is intended to test and the date at which evidence will be reviewed.
- Define in advance what would justify continuation, revision or closure.
- Reserve later capital provisionally, but do not confuse a reservation with an irrevocable commitment.
This approach does not suit every investment. Breaking an essential infrastructure replacement into fragments may increase risk and cost. Nor should staging become an excuse to begin weak projects cheaply. Its value is specific: it reduces the price of being wrong where uncertainty is genuine and information can be purchased before the whole commitment is made.
Make Stopping a Resource Decision
The hardest part of rationalisation is rarely rejecting a proposal that has not begun. It is withdrawing resources from work that remains individually defensible.
That is why stop decisions must name what is released. “Defer Project C” sounds like loss. “Release two analysts in January, one test environment in March and £1.1 million for the higher-confidence service change” reveals the portfolio benefit. The same discipline applies to combining projects: integration is worthwhile only if it removes duplicated work or resolves a dependency, not because a larger programme looks more strategic.
A portfolio office can maintain this view with a small set of artefacts: the commitment register, the capacity ledger, a dependency map, and an assumptions log. The artefacts matter less than the questions they force into the room. What have we committed? Which scarce resource is now the limiting factor? What new evidence has arrived? Which decision should change because of it?
Rationalisation Is a Leadership Test
The attraction of scoring is partly emotional. It promises that a difficult allocation can be made impersonal. Yet rebuilding after a severe downturn requires leaders to do precisely what the model cannot: distinguish obligation from opportunity, recognise when two attractive proposals cannot coexist, and withdraw support from work that remains respectable but no longer deserves scarce capacity.
A portfolio is not rational because every project has a score; it is rational when the organisation can explain what it will not start, what it will stop, and what evidence would change its mind.
The recovery will tempt many organisations to refill the pipeline as quickly as it was emptied. That would mistake activity for confidence. Real confidence is the ability to commit selectively, learn early and redirect without drama.
The old portfolio asked, “Which projects are best?” The rebuilt portfolio must ask a harder question: “Which sequence of commitments gives us the strongest position if our assumptions prove wrong?”