The most expensive decisions I have watched get made were the ones where nobody named the alternative. A management team brings a reinvestment case to the board. The numbers are credible, the payback looks reasonable, the team is confident. Everyone approves it. What nobody says out loud is that the dollar was never compared to anything — it was compared to zero. Against zero, almost every proposal wins.
Part of my work as a partner at Legacy Ventures is deciding where the next dollar goes across a set of businesses that are not equally good at using it. That job is the whole job. Underwriting, diligence, operating support — all of it is upstream of a much simpler question, asked over and over: given what we know today, what is the single best home for this dollar? A capital allocation strategy is nothing more than a repeatable, honest way of answering that question when the options are not comparable on their face.
The five places a dollar can go
At the portfolio level, the destinations are finite. That is the useful part. A dollar can be reinvested in the business that produced it. It can buy something — a competitor, a capability, a customer base. It can retire debt. It can sit on the balance sheet as insurance and optionality. Or it can leave the business entirely, which at the portfolio level does not mean disappearing; it means going to a sibling company, to a new investment, or back to the people whose money it was in the first place.
That fifth destination is the one that separates portfolio allocation from single-company allocation. An owner-operator with one business compares reinvestment against paying themselves. We compare it against every other claim in the portfolio, including claims that have not been written up yet. The dollar sitting in a company with mediocre reinvestment options is not neutral. It is actively worse than the same dollar in a company with a queue of high-return projects it cannot fund.
So the honest framing is not "is this project good?" It is "is this project better than the best thing the same dollar could do anywhere else we have visibility into?" That reframing kills a surprising number of perfectly good projects, and it should.
The next dollar, not the average dollar
Return on invested capital gets quoted as a property of a business. It is not. It is a property of a business's history. A company can have an excellent ROIC built from a founding decision made years ago and have nothing left to do with new money except add capacity nobody is asking for. Meanwhile a business with an unremarkable blended return can have three genuinely good projects it has been starving because the capital never got allocated its way.
The number that should drive reinvestment decisions is the marginal one: what does the next dollar earn, over what period, with what confidence. That number is harder to produce and easier to argue about, which is exactly why teams reach for the average instead. When a management team presents historical ROIC as justification for new spending, they are answering a question I did not ask.
The practical test I like is to force the ranking. Ask a CEO to list every project competing for capital, then ask which one they would cut first if the budget fell by a quarter. The order that comes back is far more informative than any single business case. Teams that cannot produce the order have not actually compared their options; they have simply approved them in sequence. This is where CEO capital allocation quietly separates the good from the merely competent — not in the quality of any one investment, but in the willingness to rank.
The option with a champion usually wins
Every proposal that reaches a board arrives with an advocate. The alternatives — the dollar going somewhere else, the dollar not being spent — arrive with nobody. That asymmetry is the single largest source of bad allocation I have seen, and it has nothing to do with intelligence or integrity. It is structural.
The fix is also structural. Someone has to be assigned to argue for the money leaving. Not as theatre, and not as a hostile exercise, but as a standing responsibility. When the counter-case is a named job rather than an ambient obligation, the quality of the discussion changes immediately. The proposal either survives contact with a real alternative or it does not, and either outcome is useful.
Most of this work happens outside the meeting. By the time a capital request is on an agenda, the framing is usually set and the meeting is a ratification exercise. The comparison has to be built in the weeks before, in the calls where you ask the question before anyone has invested their credibility in the answer. I have written elsewhere about what a board actually does between the meetings, and this is the highest-value version of it. A board that only sees allocation decisions in their finished form is not allocating anything. It is approving.
What a decision forecloses
Two projects with identical returns are not identical investments. One might tie up capital for six years; the other might return it in eighteen months, at which point you get to make a fresh decision with better information. That difference is enormous and it almost never appears in the comparison, because the models usually terminate at a return figure and stop.
I try to ask three things about duration before I ask about return. When does the capital come back? What can we still change once it is committed? And what does saying yes here make impossible somewhere else? The third question is the one that matters most in a portfolio, because capacity is not only financial. A management team executing a large acquisition is not simultaneously executing an operational turnaround. Attention is a scarcer input than money, and it is never on the term sheet.
This is also why the period right after a transaction is so decisive. The first allocation choices set the pattern for everything after, which is why I believe value creation starts the day after closing rather than at the first strategic offsite. The habits formed in the early months determine whether the next four years of reinvestment decisions get argued properly or rubber-stamped.
Capital released is capital allocated
Allocation is treated as a spending discipline. Half of it is a subtraction discipline. Every dollar trapped in a low-return product line, an underperforming geography, or a customer segment that consumes service capacity without paying for it is a dollar already allocated — badly, and by default.
Freeing that capital is usually harder than raising new capital, because it requires admitting a previous decision has stopped working. It also tends to be the highest-return move available, since the money is already inside the business and does not need to be found. Deciding which customers not to serve is a capital allocation decision wearing a commercial costume. So is discontinuing a product, closing a location, or ending a partnership that generates activity but not return.
The word to watch for here is "strategic." When an initiative cannot clear the hurdle on its own numbers and gets defended as strategic, that defence is sometimes right and usually not. The honest version converts the strategic claim into something specific: it opens a market we can quantify, it protects a position we can name, it buys an option we would otherwise have to pay for. If the claim cannot survive that translation, it is not a strategy. It is an attachment.
Grade the decision, not the outcome
The habit that has improved my own allocation more than any framework is embarrassingly simple: write down what you expected, what you compared it to, and what would make you wrong. Then go back and read it. Not to score wins and losses — outcomes are noisy, and a good decision can produce a bad result — but to see whether the comparison was real or whether the alternative was invented after the fact to make the preferred answer look considered.
Most portfolios do not have an allocation problem. They have a comparison problem. The projects are fine, the teams are capable, the models are competent. What is missing is a moment where two genuine options sit side by side and one of them has to lose. If we build that moment into how we work, the returns take care of themselves. If we do not, we will keep approving good ideas one at a time and wondering why the whole never adds up to the sum of the parts.