Executive summary: Capital allocation is the decision about where a company’s money goes, and over a decade it determines more about shareholder outcomes than operating performance does. Most companies allocate by budget precedent and by whoever argues most persuasively, rather than by ranking every use of capital against every other. This guide covers the full menu of uses, how to set a hurdle rate that means something, why exploiting existing assets almost always outranks buying new ones, and how to tell whether your allocation is actually working.
- What is capital allocation?
- Why does it matter more than operating performance?
- What are the uses of capital?
- Why does exploiting existing assets outrank buying new ones?
- How do you set a hurdle rate that means something?
- What is the LEAD Doctrine?
- Why allocate asymmetrically rather than evenly?
- How do you split maintenance and growth capital?
- How do you decide between buying and building?
- When should you divest?
- How do you fix capital request theater?
- What metrics tell you allocation is working?
- What are the most common capital allocation mistakes?
- Capital allocation: operator FAQ
- About the Stagnation Assassin
What is capital allocation?
Capital allocation is the deliberate ranking of every available use of a company’s money against every other, then funding in that order. It covers reinvestment, acquisitions, debt reduction, dividends, and buybacks. Done properly it is a comparative exercise, which is precisely what distinguishes it from the budgeting process most companies actually run.
The comparative part is the whole discipline. A budget asks whether each request is justified on its own terms. Capital allocation asks which of these uses produces the most value per dollar, and funds accordingly, which frequently means declining requests that would clear a standalone hurdle because something else clears it by more.
Most companies do not do this. They run a budget cycle where each business unit submits requests, finance applies a haircut, and the outcome closely resembles last year’s allocation adjusted at the margin. Nobody ever asks whether the money committed to a mature division would produce more in a growing one, because the process is not structured to make that comparison possible.
I have worked inside Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, and the organizations that treat allocation as a genuine ranking exercise behave visibly differently. They move money between businesses without ceremony. They are comfortable starving a division that cannot earn its cost of capital. And they hold cash without embarrassment when nothing clears the bar, which is a discipline most management teams find almost unbearable.
Why does it matter more than operating performance?
Because allocation decisions compound over years while operating improvements are largely annual. A business that earns twenty percent on incremental capital and reinvests heavily will outperform a better-run business that earns eight percent, regardless of how well either is operated. Over a decade the allocation choice dominates the operating one.
The arithmetic is worth walking. Two companies both generate 50 million dollars of cash annually. The first reinvests it at 20 percent returns. The second reinvests at 8 percent. Both are competently managed, both have good products, both have engaged workforces. After ten years the difference between them is not marginal, it is categorical, and no amount of operational excellence in the second company closes it.
This is uncomfortable for operators, myself included, because operating improvement is where the craft is and where the satisfaction lives. But a plant manager who improves throughput 15 percent has done excellent work that will be substantially undone if the corporation then invests the resulting cash into a business earning below its cost of capital. The operating gain is real and it is subordinate.
The practical implication is that allocation authority deserves more attention than it typically receives. In many industrial companies the capital process is administered by finance as a compliance exercise, with the actual decisions driven by whichever division president is most persistent. That is allocation by advocacy, and advocacy correlates poorly with returns.
What are the uses of capital?
There are five: reinvest in the existing business, acquire something, pay down debt, pay dividends, and repurchase shares. Every dollar goes to one of them. The discipline is ranking all five against each other continuously rather than treating reinvestment as automatic and the others as what happens to the remainder.
Reinvest in the existing business
Maintenance capital to keep operating, and growth capital to expand. This is where most industrial capital goes by default, and default is the problem. Reinvestment should compete for funding on the same terms as everything else, and a large share of what gets classified as necessary growth capital would not survive that comparison.
Acquire
Buying capability, capacity, market position, or technology. Attractive because it is fast and visible, dangerous because the price paid determines the return and the acquirer usually has less information than the seller. Discipline here means being willing to walk away frequently.
Pay down debt
Guaranteed return equal to your after-tax cost of debt, with the additional benefit of reducing fragility. Unglamorous and frequently correct, particularly for cyclical businesses where the ability to survive a downturn is itself a competitive advantage.
Pay dividends
Returning cash to owners when you cannot deploy it above your cost of capital. This should be a considered decision rather than a permanent commitment, though in practice dividend expectations become sticky enough that they function as a fixed obligation.
Repurchase shares
Sensible when the shares trade below intrinsic value and nothing else clears the hurdle, value-destroying when done at high valuations to support the share price or offset dilution. The test is price against value, and companies that buy back mechanically regardless of price are simply transferring value to exiting shareholders.
Why does exploiting existing assets outrank buying new ones?
Because capacity you already own costs nothing to acquire and is usually available in weeks rather than quarters. Most industrial assets run well below their real potential, so the first question on any capacity request should be how much output the existing asset would produce if it were fully exploited. That answer frequently eliminates the request entirely.
This is the highest-return discipline in capital allocation and it is routinely skipped, because exploitation is unglamorous and a capital request is a visible act of leadership.
Two cases from my own work make the point concretely. A foundry identified a molding operation as its limiting process and requested a 2.8 million dollar press with a 14 month lead time. Arriving three months later, I found that operation idle 35 percent of the time waiting for die changes and maintenance performed during production hours. Cutting the changeover from 73 minutes to 18 and moving all preventive maintenance into scheduled breaks raised utilization from 65 to 94 percent. The throughput gain exceeded what the new press would have added. The requisition was cancelled. Capital avoided: 2.8 million dollars. Implementation cost: 85,000 dollars.
A packaging operation faced a 40 percent demand increase and a 15 million dollar request for two new lines with an 18 month lead time. The existing lines were running at 62 percent overall equipment effectiveness, meaning 38 percent of capacity was already there and unused. Fixing changeovers, micro stoppages, and first-pass quality delivered the full 40 percent for roughly 600,000 dollars in six months. The presses were never bought and 14.4 million dollars remained available for growth.
Two capacity requests totalling $17.8M were eliminated for $685,000 of exploitation work, delivered in months rather than the 14 to 18 month equipment lead times. In both cases the capacity was already inside the existing assets. The requests were not dishonest. Nobody had asked what the current equipment would produce if it were fully used.
The rule that follows is simple and should be policy: no capacity capital request proceeds without a documented answer to what the existing asset would produce fully exploited, and what it would cost to get there. Most requests will not survive the question, and the ones that do will be genuinely necessary.
How do you set a hurdle rate that means something?
Set it above your weighted average cost of capital by a margin that reflects execution risk, then apply it consistently and decline everything below it. A hurdle rate that gets waived for strategic reasons is not a hurdle rate, it is a formality, and organizations learn within one cycle which projects need real returns and which need a good story.
The mechanics are straightforward. Establish your cost of capital honestly, add a spread for the fact that projected returns systematically exceed realized ones, and you have a number. Whether that number is 12 or 15 percent matters far less than whether it is enforced.
Enforcement is where this collapses in practice, and it collapses through a specific mechanism: the strategic exception. A project fails the hurdle, its sponsor argues that the financial analysis misses strategic value, and it gets approved anyway. Sometimes that argument is legitimate. Usually it is a way of saying the numbers do not work but we want to do it regardless.
Two disciplines keep hurdle rates real.
Require strategic exceptions to be explicit and rare. If a project is genuinely strategic and below hurdle, approve it as a named strategic exception with the shortfall quantified, recorded, and reviewed. What kills discipline is unmarked exceptions that let everyone pretend the hurdle held.
Audit realized returns against projections. Almost nobody does this and it is the single most effective corrective available. Pull every project above a threshold from three years ago and compare actual returns to what was promised. The gap is usually large, it is usually systematic, and the exercise permanently changes how the next round of projections is written.
What is the LEAD Doctrine?
The LEAD Doctrine organizes capital decisions around four tests: Legacy, whether the investment outlasts the current management team; Endurance, whether it strengthens the ability to survive a downturn; Allocation, whether it beats every competing use; and Defense, whether it protects an advantage you already hold.
Legacy
Does this build something durable, or does it produce a result inside the current planning horizon and then decay? Legacy capital creates assets, capabilities, positions, and relationships that continue producing after the people who authorized them have moved on. Much of what passes for growth investment fails this test badly.
Endurance
Does this make the company more able or less able to survive a serious downturn? Capital that raises fixed costs, adds debt service, or commits to volumes that only work at peak demand reduces endurance. In cyclical industrial businesses, the capacity to survive a bad two years is itself a competitive weapon, because it lets you buy assets from the companies that could not.
Allocation
Does this beat the alternatives, including doing nothing? This is the comparative test, and it is the one most capital processes structurally cannot perform because requests are evaluated in isolation. A project that clears the hurdle is not automatically worth funding if another clears it by more.
Defense
Does this protect a position you already have? Defensive capital is unglamorous and frequently the highest return available, because protecting an existing advantaged position usually costs far less than building a new one. Maintenance capital at a constraint is the clearest example: small spend, catastrophic consequence if skipped.
The four tests are deliberately not weighted, because their relative importance changes with circumstance. A business under liquidity pressure weights Endurance heavily. A business with an unassailable position and surplus cash weights Legacy. What the doctrine enforces is that all four questions get asked, which prevents the common failure of evaluating capital purely on projected return while ignoring what it does to fragility.
Why allocate asymmetrically rather than evenly?
Because returns are not evenly distributed, so even allocation guarantees you underfund your best opportunities and overfund your worst. Concentrating capital on the small number of positions with genuinely superior returns produces better outcomes than spreading it, even though spreading feels prudent and is politically far easier.
Even allocation is what organizations default to in the absence of conviction. Every division gets roughly its historical share adjusted for inflation. It feels fair, it minimizes internal conflict, and it is a reliable way to produce mediocre returns, because it treats a business earning 25 percent on incremental capital identically to one earning 6 percent.
The reason even allocation persists is not analytical. It is that asymmetric allocation requires telling capable executives running decent businesses that their capital is going elsewhere. That conversation is genuinely difficult, and avoiding it is expensive in a way that never appears on any report.
Practically, asymmetric allocation means a small number of positions receive materially more than their proportional share, most receive maintenance capital only, and some receive nothing while being managed for cash. The last category is the hardest to sustain, because a business managed for cash will underperform on growth metrics and its leadership will correctly point that out every quarter.
The connection to portfolio and operational work is direct. If a small fraction of your products and customers generates most of your profit, and if one process limits your output, then the capital that matters concentrates around those. Even allocation is the financial expression of the same error as spreading improvement effort evenly across a plant: it is what an organization does when it has not identified where the leverage is.
How do you split maintenance and growth capital?
Separate them explicitly and hold them to different tests. Maintenance capital preserves existing earning capacity and should be evaluated on consequence of failure rather than on return. Growth capital creates new earning capacity and must clear the hurdle. Blending the two is how companies quietly starve maintenance to fund growth stories.
The blending problem is worth being specific about, because it is common and its damage is delayed. When both categories compete in one pool, growth projects with attractive projected returns systematically beat maintenance projects with no return at all, since maintenance merely preserves what already exists. Do this for several years and you have a company with impressive growth investments running on deteriorating infrastructure.
Maintenance capital should be evaluated on a different question entirely: what happens if we do not do this, and when? A maintenance item at your constraint has a consequence measured in system throughput, not in project return. Deferring it saves a small amount of capital and risks an amount of output that dwarfs it.
The placement discipline matters here as much as it does in maintenance strategy generally. Maintenance capital should follow constraint status rather than asset value. Plants routinely allocate their maintenance budgets toward their newest and most expensive equipment while the old asset that actually governs output runs to failure. That is backwards, and it is one of the most reliably expensive misallocations in industrial operations.
How do you decide between buying and building?
Compare total cost including time, not just capital outlay. Acquisition buys speed and existing capability at a price set by a seller with better information. Building costs less in cash and more in time and execution risk. The deciding variable is usually how much the time difference is worth, which most analyses fail to quantify at all.
The standard comparison puts acquisition cost against build cost and picks the cheaper. That comparison is incomplete in both directions.
What acquisition really costs. The purchase price, plus integration cost, plus the failure rate of integrations, plus the information asymmetry that means the seller knows what they are selling better than you do. Acquisitions of operating businesses also import an operating culture, and the cost of changing one is routinely underestimated.
What building really costs. The capital, plus the time to competence, plus the opportunity cost of the management attention consumed, plus the risk that you build something the market has moved past. Building looks cheap on a spreadsheet because the time and attention costs sit outside the model.
What the time difference is worth. This is the missing variable. If the capability generates value from the day it exists, then eighteen months of difference has a calculable price, and that price frequently exceeds the entire gap between the two options. Quantifying it converts a strategic debate into an arithmetic one.
One structural preference worth stating. Where the capability in question is close to what you already do, building usually wins, because your execution risk is low and you avoid the acquisition premium. Where the capability is genuinely distant from your competence, buying usually wins, because your build estimate is almost certainly optimistic in ways you cannot see from inside.
When should you divest?
Divest when a business cannot earn its cost of capital under any plan you actually believe, when it consumes disproportionate management attention relative to contribution, or when it is worth more to another owner than to you. The last test is the one companies apply least and it is frequently the most compelling.
Divestiture is the least emotionally available capital decision, which is why underperforming businesses persist for years past the point of obvious diagnosis. The resistance is rarely financial. It is that selling a business is publicly legible as a failure, whereas continuing to run it poorly is not.
Three tests that clarify the decision:
The honest plan test. Not whether a plan exists, but whether you believe it. Most persistently underperforming units have a credible-looking turnaround plan that has been credible-looking for several years running. Ask what specifically will be different this time and whether the answer would convince an outside investor.
The attention test. How much senior management time does this business consume relative to its contribution? Attention is the scarcest corporate resource, and a small troubled business can consume an amount grossly disproportionate to any plausible outcome.
The better-owner test. Is there an owner for whom this business is worth substantially more, because of adjacency, scale, or capability you lack? If so, the value gap between what it is worth to them and to you is real money, and capturing it is a legitimate allocation decision rather than an admission of defeat.
How do you fix capital request theater?
Require every request to state what happens if it is declined, audit realized returns against projections publicly, and make the comparison between competing requests explicit rather than evaluating each in isolation. Projections improve rapidly once people know they will be checked against outcomes.
Capital request theater has a recognizable signature. Projections cluster just above the hurdle rate. Every request is described as essential. Alternatives are presented in a form designed to be rejected. And nobody ever revisits whether the last round delivered what it promised.
Four corrections work.
Require the do-nothing case. Every request states specifically what happens if it is declined, in numbers. This kills the essential framing immediately, because a genuinely essential project has a specific and severe consequence of decline, and most do not.
Audit realized returns. Publicly, annually, against the original projections. This single practice does more to improve forecast honesty than any amount of process design, because it changes the incentive of the person writing the projection.
Force explicit comparison. Present requests as a ranked list competing for a fixed pool rather than as individual approvals. The moment sponsors know their project is competing against a specific alternative rather than against a hurdle rate, the quality of the cases improves markedly.
Attach the cost of delay to the decision itself. For projects that genuinely earn strong returns, every month in review is a real cost. Making that number visible compresses review cycles and stops good projects from dying in committee. It also works in reverse, exposing that many projects framed as urgent have almost no cost of delay at all, which is diagnostic.
What metrics tell you allocation is working?
Track return on incremental invested capital rather than overall ROIC, the ratio of realized to projected returns on completed projects, the share of capital going to your highest-return positions, and how often capital moves between businesses. Overall ROIC is dominated by legacy investments and can look healthy while current allocation is poor.
Return on incremental invested capital
The return generated by capital deployed recently, isolated from the historical base. This is the number that reflects decisions the current team actually made, and it can be substantially worse than headline ROIC while total returns still look acceptable, which is exactly the situation that destroys value slowly.
Realized versus projected returns
The ratio, tracked by sponsor and by business unit. A consistent pattern of projections exceeding outcomes is a forecasting culture problem rather than a series of individual misses, and it is correctable once measured.
Concentration of capital
What share went to your highest-return positions versus being spread evenly. A capital allocation that closely resembles a proportional distribution of revenue is a signal that no ranking is actually occurring.
Capital mobility
How frequently and how much capital moves between businesses. Organizations where every unit receives approximately its historical share year after year are not allocating, they are budgeting, whatever the process is called.
Overall return on invested capital is dominated by decisions made years ago and can look healthy while current allocation is poor. Return on incremental invested capital isolates the decisions the present team actually made. When the two diverge persistently, the company is living on legacy investments while quietly destroying value with new ones.
What are the most common capital allocation mistakes?
Five recur: treating reinvestment as automatic rather than competitive, adding capacity where the business is not constrained, allocating evenly to avoid internal conflict, waiving hurdle rates for strategic reasons without recording the exception, and never auditing whether past projects delivered.
Mistake 1: automatic reinvestment
Capital flows to the existing business by default because that is where the requests originate. Reinvestment should compete against acquisitions, debt reduction, and returning cash on identical terms. A large share of routine growth capital would not survive that comparison, and it is never asked to.
Mistake 2: capacity where you are not constrained
The most expensive operational misallocation. Adding capacity to a process that was never limiting output produces a better utilization number and no additional throughput. Confirm what actually limits output before funding any capacity request, and confirm it with measurement rather than with the requesting manager’s assessment.
Mistake 3: even allocation to keep the peace
Distributing capital proportionally avoids difficult conversations with capable executives and guarantees mediocre aggregate returns. The conversation is the job. Avoiding it transfers the cost to shareholders in a form nobody will ever attribute to the decision.
Mistake 4: unrecorded strategic exceptions
A hurdle rate that is waived without the waiver being named, quantified, and reviewed teaches the organization that the hurdle is negotiable. Within two cycles every marginal project arrives wrapped in strategic language.
Mistake 5: never looking back
Companies conduct elaborate analysis before committing capital and almost none afterwards. Without a realized-return audit there is no feedback loop, which means projection quality never improves and the same optimistic assumptions recur indefinitely.
The mistake I have made personally was approving capacity investment before confirming where output was genuinely limited. The analysis was competent, the projections were reasonable, and the capacity was added to a process that was not the constraint. Utilization improved at that operation, the system produced no more than before, and the capital was permanently committed. It taught me to treat the constraint question as a gate on every capacity request rather than as an operational detail to be settled later.
Capital allocation: operator FAQ
What is capital allocation?
The deliberate ranking of every available use of a company’s money against every other, then funding in that order. The five uses are reinvestment, acquisition, debt reduction, dividends, and share repurchases. What distinguishes it from budgeting is that it is comparative, so a project can clear a hurdle and still be declined.
How do you set a hurdle rate?
Set it above your weighted average cost of capital by a margin reflecting execution risk, then enforce it consistently. The specific number matters far less than whether it holds. Strategic exceptions should be explicit, quantified, and rare, since unmarked exceptions teach the organization that the hurdle is negotiable.
Should you buy new equipment or improve existing assets first?
Improve existing assets first, almost without exception. Most industrial equipment runs well below its real potential, and exploitation delivers in weeks what equipment delivers in quarters. Require every capacity request to document what the existing asset would produce fully exploited and what reaching that would cost.
Why does capital allocation matter more than operating performance?
Because allocation decisions compound over years while operating improvements are largely annual. A business reinvesting at twenty percent returns will outperform a better-operated business reinvesting at eight percent, regardless of execution quality. Over a decade the allocation choice dominates the operating one.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B+ in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as VP of Global Product Strategy. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via the speaking page or connect with him on LinkedIn.
Next step: test one capital request
Take the largest capacity request currently sitting in your approval process and ask one question: what would the existing asset produce if it were fully exploited? Book a 20 minute review and I will help you answer it. In my experience most requests do not survive the question, and the ones that do are worth funding fast. Start the review here.