Diminishing Returns

011 min

Diminishing Returns at a glance

• What it is: Adding more of one input to a setup that is otherwise fixed eventually makes each new unit less productive than the last.
• The number that matters: Marginal product — the extra output from the last unit added — not total output, which can keep rising even as each new unit adds less.
• Where it breaks: People confuse diminishing returns (one input, short run) with diseconomies of scale (every input, long run). They are different mechanisms with different fixes.

021 min

The problem Diminishing Returns names

A startup falling behind schedule often reaches for the same fix: hire more people. A support team doubles its headcount hoping to clear a backlog twice as fast. It rarely works out that way. New hires need training before they help at all, existing staff lose time explaining the product to them, and the queue of unresolved tickets keeps growing while the team adjusts. The mistake is treating headcount as if it scales output in a straight line, when the fixed parts of the operation — one shared ticketing system, one onboarding process, one manager who can review only so many cases a day — cap how much each additional hire can actually add. Ignoring that cap is expensive: money spent on wages that return progressively less, and sometimes a backlog that gets worse before it gets better.

032 min

Understanding Diminishing Returns

Diminishing returns describes what happens to output when one input to a production process keeps increasing while every other input is held fixed. Economists call the extra output from one more unit of that input its marginal product — the amount total output rises by when a single unit is added, everything else unchanged. The law says marginal product rises at first, then falls, as the variable input keeps growing against a setup that does not grow with it.

The cause is the fixed inputs themselves. A factory floor, a warehouse, a machine, or a plot of land does not expand just because more workers or more fertilizer arrives. Early units of the variable input use spare capacity the fixed inputs already have: an idle machine gets a second operator, empty floor space gets a new workstation, soil that was under-fertilized finally gets enough nutrients. Each of these early additions raises output by close to what the one before it did, sometimes more, because the fixed setup was not yet fully used. Once the fixed inputs are fully occupied, that spare capacity runs out. A third worker at one machine has to wait a turn; a crowded floor slows everyone down; soil that already holds enough nutrients gains almost nothing from more fertilizer, and eventually the fertilizer itself does harm. Output kept rising with each addition, but by smaller amounts — this is diminishing returns. If the variable input keeps growing past that point, marginal product can turn negative, so total output falls with the next unit added; economists call this negative returns, and the same underlying cause — a fixed factor running out of room — produces it.

The law holds only in the short run, defined here by economics rather than the calendar: the period in which at least one input to the process cannot be changed. That same split between what is fixed and what can move is the one behind fixed and variable costs in accounting. In the long run every input is variable, including the machine, the floor space, and the land itself, and a different set of ideas — returns to scale — governs what happens when all of them grow together.

041 min

Where Diminishing Returns comes from

The law traces to Anne Robert Jacques Turgot, a French economist and finance minister who worked a century before the term existed. He is thought to have been the first political economist to have postulated something like the law of diminishing marginal returns in agriculture. Turgot argued that adding more labor and capital to a fixed plot of land does not raise the harvest in proportion: each increase [in an input] would be less and less productive. His best-known work, Reflections on the Formation and Distribution of Wealth, was written in 1766 and published in 1769-1770; some historians date his specific observation about diminishing returns to a separate 1767 essay on taxation, so the precise year is not settled to the month.

Turgot's insight was picked up formally decades later. In 1815, David Ricardo, Thomas Malthus, Edward West, and Robert Torrens each applied the concept of diminishing returns to land rent, during a parliamentary inquiry in England into why grain prices were so high. They concluded that the prices of grain had risen because of the Napoleonic Wars, which disrupted international trade and pushed farmers onto undeveloped, more distant land.

051 min

The numbers

Take a small bakery kitchen with one oven and one prep counter — a hypothetical case, with round numbers chosen only to make the arithmetic easy to follow. The oven and counter are the fixed inputs; the number of bakers working a shift is the variable one.

The first baker works alone and produces 20 loaves. The second does not double that; output rises to 46, an extra 26 loaves, because two people can now use the oven and the counter at the same time without waiting on each other. From the third baker onward, the extra loaves from each new hire keep shrinking — 22, then 16, then 6, then 1 — because the kitchen has only one oven and one counter, and a sixth baker is mostly standing in the doorway. Total output never falls in this hypothetical case, but the return on each new hire does, and by the sixth baker it is barely worth the wage paid for the shift.

061 min

A second case

The bakery stays polite: output keeps rising, just by less each time. Software staffing shows the same mechanism taken further, to the point where an extra unit of input subtracts instead of adds. Fred Brooks was best known for managing development of IBM's System/360 family of mainframe computers and the OS/360 software support package, then later writing candidly about those experiences in his seminal book The Mythical Man-Month. That book states what is now called Brooks's law: adding manpower to a late software project makes it later.

Brooks's own explanation is a direct case of diminishing, then negative, returns. It takes some time for the people added to a project to become productive, because new engineers first have to learn a codebase that existing engineers must stop and explain to them, temporarily reducing the team's own output while training happens. And communication overhead increases as the number of people increases, since every new engineer adds another person everyone else has to stay in sync with. Early hires on an under-staffed project add real capacity, the way the second baker did. Past a certain team size, the training cost and the coordination cost of each new engineer outweigh the code they personally ship, and the marginal return on one more hire turns negative — the project finishes later, not sooner, for having grown.

072 min

What this means for your work

A PM staffing a rescue effort on a slipping launch date faces the mechanism directly: the honest question is not whether more people would help, but how many more the existing tech leads can onboard and review before the marginal engineer's output falls to zero. Adding two people who can start immediately, with no ramp-up, is usually worth it; adding six people who all need the same two senior engineers to explain the codebase to them usually is not.

An engineer scaling a service under load meets the same pattern from the infrastructure side. Doubling the number of application servers behind a database that is already the bottleneck adds very little throughput, because the database's own capacity is the input that is actually fixed, not the number of servers asking it for data. Improving the service's real efficiency means finding that one fixed limit and widening it, not adding more of an input that already has room to spare.

A founder deciding how much to spend on paid acquisition meets it from the marketing side: the first thousand dollars of ad spend usually reaches the people most ready to buy, and each additional thousand reaches people who take more convincing, so the cost of winning one more customer climbs even while total sign-ups keep growing. The decision this forces is not whether to keep spending, but at what spend the next dollar's return drops below what it costs to raise it.

081 min

How to apply Diminishing Returns

The concrete check is to track marginal product, not total output. For any input added in discrete units — people, ad spend, servers, machines — measure what the last unit alone added, not what the whole group produced together. A team that grew from eight people to ten and shipped five percent more work added two people for a five percent gain; that marginal number, not the team's total output, is what says whether an eleventh hire is worth it.

Where the input is money, run the comparison in the same currency: does the next unit of spend return more than it costs to raise. A proper cost-benefit analysis of just the next unit, tested on a small, reversible slice of budget rather than committed in full, gives that marginal number directly instead of a guess.

Where a single fixed input is the real limit — one database, one review queue, one approval step — fix that first. Adding more of the variable input around a fixed constraint is the least effective way to raise output; widening the constraint itself, even briefly, usually beats it.

091 min

When Diminishing Returns breaks

The most common misuse is treating diminishing returns as a reason to stop adding an input the moment marginal product starts falling, rather than the moment it falls below what that unit costs. Marginal product can decline for several units in a row and each of those units can still be worth adding — a fourth baker who adds only sixteen extra loaves is still profitable if a baker's wage costs less than sixteen loaves are worth. Cutting an input the instant its marginal return starts falling, instead of when it turns unprofitable, leaves real output on the table.

The second failure is mistaking diminishing returns for diseconomies of scale and reaching for the wrong fix. Diminishing returns is a short-run problem with one fixed input; the fix is removing that specific bottleneck. Diseconomies of scale is a long-run problem where growing every input at once makes coordination itself more expensive, and no single bottleneck removal solves it. A company that responds to diminishing returns by shrinking the whole organization, instead of fixing the one constrained resource, treats the wrong disease.

101 min

Diminishing Returns vs. nearby concepts

Diminishing returns is confused most often with diseconomies of scale, because both describe output growing more slowly than input. The deciding fact is which inputs are allowed to change. Diminishing returns fixes every input except one and blames the fixed input for the slowdown. Diseconomies of scale lets every input grow together and blames coordination itself: more people means more meetings and more distance between a decision and the person who acts on it. A compounding process runs the opposite way, each unit adding more than the last, which is why the two are useful to hold side by side.

112 min

Frequently asked questions about Diminishing Returns

?7 questions

Questions people ask

§6 sources

Sources