A budget is a strategy with a price tag. Whatever you say your priorities are, the real ones are the line items you actually fund, and for most organisations those line items barely move from one year to the next. Portfolio management is the work of changing that on purpose: looking at every investment as one set of competing claims on finite money and attention, then deciding, deliberately, where the next dollar does the most good.

The quick version

  • Manage the set, not the project. Prioritisation is comparing your investments against each other for the same scarce money, not approving each one on its own merits.
  • Inertia is the default, and it's expensive. Capital tends to land where it landed last year. Organisations that actively reallocate earn meaningfully higher returns over time.
  • Rank by value per unit of cost, not by who argued hardest. A simple "value ÷ effort" score, applied consistently, beats the loudest voice in the room.
  • Fund across horizons. Protect today's core, build tomorrow's growth, and seed a few longer-shot options, so you don't mortgage the future to make this quarter.

The idea in depth

The first move in portfolio thinking is older than most of the language around it. In 1970, Bruce Henderson of the Boston Consulting Group published a short essay called The Product Portfolio, arguing that a company's businesses should be seen not as a list of things to run but as a balanced set: some throwing off cash, some consuming it to grow, some to be quietly wound down. That essay became the growth-share matrix, the famous grid of stars, cash cows, question marks and dogs. The mechanics matter less than the mental shift it forced: stop judging each investment in isolation, and start asking how the pieces fund and balance one another.

Draw the boundary around the whole set before you judge any one item. The question that funds good portfolios is never "is this a good project?", almost everything on the table is a good project, defended by a smart person with a real case. The useful question is "is this a better use of the next dollar than everything else competing for it?" That reframing is uncomfortable: it means a genuinely good idea can be the right thing to not fund this year.

Be honest about where it has dated, though: the original matrix has aged unevenly. Its core assumption, that high market share reliably produces profit through scale and the experience curve, does not hold across every industry, and critics have said so since the 1970s. Treat it as a lens that prompts the portfolio question, not a formula that answers it. Market share is one signal among many, not a verdict.

Inertia is the enemy, and it has a price

The most useful research in this area is not about which framework to use, it's about how rarely leaders use any of them. In a 2012 study, McKinsey's Stephen Hall, Dan Lovallo and Reinier Musters tracked how more than 1,600 US companies allocated capital across their business units over fifteen years. The standout finding was how little the numbers moved: the amount of money a given unit received from one year to the next had a correlation of 0.92, in plain terms, this year's budget is almost entirely predicted by last year's. Strategy decks changed; the money mostly stayed put.

"Companies that reallocated more resources… earned, on average, 30 percent higher total returns to shareholders annually than companies that reallocated less."

That inertia turned out to be costly. The third of companies that reallocated the most, shifting an average of 56% of capital across units over the period, delivered roughly 30% higher total returns to shareholders per year than the third that reallocated least, a gap McKinsey reported as surprisingly consistent across sectors (Hall, Lovallo & Musters, McKinsey Quarterly, 2012). The active reallocators were also more likely to survive, less likely to be acquired or go bankrupt over the window.

The practical fix is to make reallocation a scheduled act rather than an accident. Pick a cadence, quarterly for fast-moving teams, and at each one ask a deliberately blunt question: if we were funding from a blank sheet today, would this still get money? The blank sheet breaks the gravitational pull of last year's budget. Why does inertia win by default? Because cutting funds someone always feels like a verdict on them, and protecting last year's allocation avoids the fight. Naming that dynamic out loud is half the cure.

One caveat, because the finding gets misread: correlation is not a licence to churn. Yanking funding before a bet has had time to prove out destroys value as surely as freezing it. The finding argues for willingness to move money, governed by evidence, not restlessness.

flowchart TD
    A(["Every candidate investment
competing for the same money"]) --> B(["Score each: value ÷ cost-to-deliver"])
    B --> C(["Rank highest score first"])
    C --> D(["Draw the funding line
where the budget runs out"])
    D --> E(["Above the line:
fund now"])
    D --> F(["Below the line:
defer, shrink, or kill"])
    E --> G(["Review on a fixed cadence
re-rank, reallocate"])
    F --> G
    G --> A
Prioritisation as a loop, not a one-off approval queue. Leaders Loop

Rank by value per unit of cost

If reallocation is the when, the how needs a rule that survives a room full of advocates. The most durable rule is also the simplest: rank work by the value it delivers divided by the cost of delivering it, and fund down the list until the money runs out. In product and engineering circles this is formalised as Weighted Shortest Job First (WSJF), cost of delay divided by job size, a method Don Reinertsen set out in The Principles of Product Development Flow (2009) and which the Scaled Agile Framework later adopted. Reinertsen's sharpest line is the one to remember: if you only quantify one thing, quantify the cost of delay.

In practice that means forcing two numbers for every candidate before you compare them: roughly how much value does it create (and how urgent is that value, what does waiting cost?), and roughly how big is the effort. You do not need a finance team or a perfect model; relative estimates on a 1–10 scale are enough to surface the obvious truth that gets lost in debate, that a small job worth a lot should beat a big job worth a little, even when the big job has a louder champion. The discipline isn't the maths. It's that the same rule applies to everyone's pet project.

One thing this method won't do for you: a value-÷-cost score is a sorting aid, not an oracle. It rewards the quantifiable and can quietly starve the strategic-but-fuzzy, the platform investment, the capability you'll need in three years. That's exactly why ranking sits inside a horizon view rather than replacing it: you protect a slice of the budget for longer bets before you let the score sort the rest.

A worked example

The figures below are illustrative, invented to show the method, not drawn from a real company.

Imagine a mid-sized software business with roughly £4m of discretionary investment for the year and six candidates on the table, each with a credible sponsor. Finance has, sensibly, asked every sponsor for a value estimate (1–10) and a delivery-cost estimate, and computed a simple score of value ÷ cost.

flowchart LR
    subgraph H1["Horizon 1 · protect the core (~70%)"]
      A(["Checkout reliability fix
value 9 · cost £0.6m · score 15.0"])
      B(["Billing automation
value 7 · cost £0.8m · score 8.8"])
    end
    subgraph H2["Horizon 2 · build the next engine (~20%)"]
      C(["New SMB product line
value 8 · cost £1.2m · score 6.7"])
      D(["Partner integrations
value 6 · cost £1.0m · score 6.0"])
    end
    subgraph H3["Horizon 3 · seed options (~10%)"]
      E(["AI advisory feature
value 7 · cost £1.4m · score 5.0"])
      F(["Brand refresh
value 4 · cost £1.5m · score 2.7"])
    end
Illustrative figures only. Rank within a horizon allocation, not across the whole list blindly. Leaders Loop

Sorted purely by score, the order is: the checkout fix (15.0), billing (8.8), the SMB product line (6.7), partner integrations (6.0), the AI feature (5.0), then the brand refresh (2.7). A naïve team funds straight down that list until the £4m runs out, which would buy the first four and stop, leaving nothing for the future engine the business will need when today's core matures.

The portfolio-literate team does something subtler. It first splits the budget across horizons, a rough 70 / 20 / 10 between protecting the core, building the next business, and seeding longer-term options, the split Baghai, Coley and White popularised in The Alchemy of Growth (1999) and which McKinsey still publishes as an enduring idea. Then it ranks within each horizon. The checkout fix and billing automation win Horizon 1 on raw score; the SMB product line takes most of the Horizon 2 pot; and a protected ~10% keeps the AI feature alive as a small, time-boxed experiment, not because it scored well today, but because it buys an option on a future the business can't yet price. The brand refresh, which sponsors will fight for, loses on both counts and is deferred. The score did the sorting; the horizon split made sure the future got a seat at the table.

Frequently asked questions

Isn't this just budgeting with extra steps?

Budgeting allocates money once and defends it. Portfolio management treats every allocation as provisional and re-competes it on a cadence. The difference shows up in the McKinsey 0.92 figure: budgeting alone produces a portfolio that barely changes year to year, which is precisely the pattern active reallocators beat.

How do we estimate "value" when we genuinely don't know?

Use relative, not absolute, estimates, rank candidates against each other on a 1–10 scale rather than forecasting exact returns nobody believes. The goal is a defensible ordering, not a spurious number. Where uncertainty is the whole point (a Horizon 3 bet), fund a cheap experiment to buy information, then re-score once you know more.

Won't a scoring formula just get gamed by whoever inflates their numbers?

It will, if the scores are private. The fix is to make estimates public and comparable: everyone sees everyone's value and cost, side by side, in the same review. Gaming is much harder when a sponsor has to defend why their job is worth three times another in front of the people whose work they'd be displacing.

What about the strategic project that a score will always starve?

Ring-fence it before you sort. The horizon split exists for exactly this: you decide what share of the budget protects long-term and capability bets, fund those first, then let the value-÷-cost ranking sort the rest. A score should never get to veto strategy, it sorts within the space strategy has already carved out.

How often should we re-prioritise?

As often as your environment changes materially, and no less than annually. Quarterly suits fast-moving product organisations; slower-moving capital-intensive ones may run half-yearly. The trap is treating any cadence as sacred, the calendar serves the reallocation, not the other way around.

Related in the Toolkit

Where to go next