The 3 Types of Compounding Machines. One of Them Is a Trap.

The three fates of compounding machines

Start with three numbers.

Over three years, a company's operating profit rises from $1.0 billion to $1.4 billion. Invested capital rises from $5.0 billion to $9.0 billion. Incremental profit is $0.4 billion. Incremental capital is $4.0 billion. Marginal ROIC is 10%. Over the same period, FCF / Net Income falls from 85% to 55%. The market gives the company 30x FCF.

Those numbers break the label. Profit still grows. Capital returns are falling. Cash conversion is weakening. The market still trades it like a compounding machine.

The central question is not whether the company is good. It is what kind of compounding machine it is, and whether it is still that machine today.

Three mental models matter:

1. The endogenous compounder. The business itself is the snowball.

2. The allocator. The engine sits in capital allocation, not in the factory.

3. The cash cow and the former compounder. The same coin, depending on how the market prices it.

These three models form a coordinate system for long-term value.

The endogenous compounder

The endogenous compounder is the ideal machine. Capital compounds inside the business. Every extra dollar of incremental capital produces stable or expanding marginal profit. Growth does not consume increasing amounts of cash. Free cash flow expands alongside revenue. The business does the work, not a clever CEO moving money around.

Three tests matter.

First, marginal ROIC is stable or rising. This is the pulse of the compounding machine. Absolute ROIC is a record of the past. Marginal ROIC decides the future. If profit still grows but each new dollar of capital produces less profit, the engine is wearing down.

Second, free cash flow expands with revenue. An endogenous compounder does not devour cash to grow. Revenue rises, FCF rises. For mature software companies, FCF / Net Income should usually exceed 80%. If revenue grows quickly but FCF stays low and the company keeps raising money, it is a capital-hungry business, not an endogenous compounder.

Third, the reinvestment runway is long. The company can redeploy most of its profit into high-return opportunities. The longer the runway, the longer the compounding lasts. Once the runway ends, the endogenous compounder becomes a cash cow, or worse, a former compounder if management keeps investing badly.

The moat matters, but it is not static. The framework forces a forward-looking question: can this moat withstand the main erosion forces of the next decade? Signs that a moat has entered a defensive phase:

• Customer retention costs are rising.

• R&D spending increases but only maintains feature parity.

• Discounts are needed to keep customers.

• Brand spending rises without market share gains.

The company is not expanding the moat. It is paying to hold it.

The most dangerous situation is this: absolute ROIC is still high, but marginal ROIC has been falling for three consecutive years. The market sees high ROIC and keeps paying a high multiple. Falling marginal ROIC is the first alarm that the compounding engine is rusting. Next comes FCF / Net Income below 60%, and then large non-core acquisitions by management.

At that point, the company may already be a former compounder. The market is still pricing it with the old label.

The allocator

The allocator is easy to misread, and it rewards close attention. A capital allocation compounder works like this: the operating business itself cannot endlessly replicate high returns, but the organization can systematically redeploy the cash it generates into higher-return opportunities. Compounding happens at the balance sheet and the capital allocation meeting, not in the factory. The engine is management's capital allocation discipline, not the inertial growth of the operating units.

Allocators come in two forms, and their fates are completely different.

The first is the system-driven allocator. The allocation ability is institutionalized. Examples include insurance float investment mechanisms, subsidiary dividend upstreaming, a central capital allocation committee, and long-term return scorecards. If the CEO changes, the machine still runs. This is the most valuable kind of allocator. Its compounding does not depend on a genius. It depends on a system.

The second is the person-driven allocator. The allocation ability is attached to one brilliant CEO or chief capital allocator. When that person is present, the company turns lead into gold. When that person leaves, the company can quickly turn from excellent capital allocator into empire builder: expensive acquisitions, low-return diversification, scale for scale's sake. History is full of companies that went from excellent allocator to empire builder. The stock underperformed for years. The cause was the allocation engine changing souls, not operating deterioration.

Four tests separate an allocator with discipline from a pretender.

First, cross-segment capital transfer. Does the company systematically move capital from lower-return segments to higher-return segments? Calculate segment CROCI. Watch capital allocation shifts over five years. If low-return segments keep getting money while high-return segments are starved, allocation discipline is weak.

Second, acquisition ROI. For acquisitive allocators, track cumulative goodwill and intangible assets relative to cumulative operating income growth. If goodwill grows faster than operating income, acquisitions are destroying value.

Third, buyback and dividend discipline. When internal high-return opportunities are scarce, does the company return capital? Or does it invest in low-return projects out of empire-building impulse? An allocator with discipline knows when to give money back.

Fourth, reinvestment rate and marginal ROIC. If the company keeps reinvesting but marginal ROIC is below the cost of capital, it is not compounding. It is destroying value.

The market often makes one mistake: it values an allocator as if it were an endogenous compounder. An allocator can be great, but its valuation logic is different. An endogenous compounder can earn a growth premium because the business compounds automatically. An allocator must prove that its allocation ability is system-driven, not person-dependent. If allocation is person-dependent, the valuation must be discounted, and the risk must be listed as top-tier. If the key person leaves, the machine may stop.

The allocator's compounding lives in the ability to pull cash out of low-return mud and put it into high-return soil, not in the growth inertia of the operating units. That ability must be institutionalized to deserve a long-term premium.

The cash cow and the former compounder

The third mental model is one coin with two sides: the brand cash cow and the deteriorating former compounder. Both were once some kind of compounding machine. Their fate depends on how the market prices them, and what management does.

The brand cash cow has a strong brand, high margins, and very stable cash flow. But it is limited by physical capacity, scarcity maintenance, market saturation, or regulation. It has no high-return reinvestment space. It produces far more cash than it can reinvest at high returns. So its value does not come from growth. It comes from dividend and buyback discipline.

The framework gives an iron rule: a cash cow must be valued on cash returns. FCF yield above 5% is attractive. Below 3% is a red flag, because the market may be mispricing a cash cow as a growth compounder. If the market gives a low-growth, high-payout cash cow 30x FCF, that is speculation on a growth re-acceleration that may never come.

The former compounder was once an endogenous compounder, or once an excellent allocator. But its compounding engine has rusted. The danger is that the market still prices it with compounding-machine multiples.

Signals of a former compounder:

• Marginal ROIC has fallen for three or more years, while absolute ROIC is still high.

• FCF / Net Income has fallen below 60%, with no clear temporary explanation.

• Management begins large non-core acquisitions without a clear compounding logic.

• Growth requires more and more capital to maintain, meaning spending more just to stand still.

• The moat narrative is unchanged, but market share, pricing power, and customer acquisition costs are deteriorating.

If these signals appear together, the company may already be a former compounding machine. The market's biggest error is not missing a good company. It is continuing to pay a compounding multiple for a former compounder.

The framework gives a hard trigger. If a company was classified as an endogenous compounder, but any two of the following happen at the same time, marginal ROIC falls for four consecutive quarters, FCF / Net Income falls below 60%, or management begins large non-core M&A, that is the classic signature of a former compounder in denial. Sell regardless of price. The trigger is not about imminent bankruptcy. The compounding engine no longer supports the current multiple. Holding it means risking permanent capital loss.

How the three models work together

Put the three models together, and you get a decision map.

Four questions matter.

First, for every extra dollar invested, how much profit can it still earn? Look at marginal ROIC. Rising or flat suggests a possible endogenous compounder. Falling but absolute ROIC still high suggests a possible early-stage former compounder. Dependent on capital reallocation suggests a possible allocator.

Second, how much capital does growth consume? Look at capital intensity, Capex plus SBC as a percentage of revenue, and FCF conversion. FCF grows with profit in an endogenous compounder. FCF persistently below net income and constant fundraising means capital-hungry, not endogenous. Cash-rich but limited reinvestment space means a cash cow.

Third, over the next decade, how large is the reinvestment space? Look at TAM, market saturation, regulatory ceilings, physical capacity. A long runway suggests an endogenous compounder or excellent allocator. A short runway suggests a cash cow. A short runway but management insists on growth suggests a potential former compounder.

Fourth, if the CEO changes tomorrow, does the machine still run? The endogenous compounder depends on systems. The allocator depends on institutions. The cash cow depends on discipline. The former compounder depends on the market's illusion. If the entire compounding logic depends on one person, it is fragile. This is especially true for allocators.

Before trusting any label, calculate three numbers: marginal ROIC, FCF / Net Income, and goodwill growth minus operating profit growth. These three numbers together are more reliable than any story.

Run the mean-reversion stress test. Assume the business performs exactly as expected. Assume the valuation multiple reverts to its historical median. Calculate the annualized return. If it is below the risk-free rate plus 3%, the price already embeds a perfect future. For cash cows, the iron rule is stricter. Is FCF yield above 5%? Is the P/E consistent with a low-growth, high-payout profile?

Continuous verification

Companies are alive. An endogenous compounder can deteriorate into a former compounder. An allocator can shift from system-driven to person-dependent. A cash cow can be mispriced by the market as a growth stock. A former compounder can re-ignite its engine through a successful transition.

Finding a compounding machine and holding it forever is not the work. The work is to keep asking four questions. Is it still that machine? Which type is it now? Is the engine still running? Does today's price assume a machine that never wears out?

The next time you read an earnings report, calculate three numbers first: marginal ROIC, FCF / Net Income, and goodwill growth minus operating profit growth. Then apply the label. Labels are cheap. Machines are not.