Decisions are made at the margin. The relevant question is not "Is this good or bad in total?" but "What happens if I do one more unit, or one less?" Marginal thinking compares the incremental cost of the next unit to the incremental benefit. If marginal benefit exceeds marginal cost, do more. If marginal cost exceeds marginal benefit, do less. The optimum is where they are equal. Averages and totals obscure this; they mix in sunk cost and fixed cost that should not affect the next choice.
The margin is where you have agency. Past spending, past output, and fixed commitments are done. The only lever is the next action. A factory that has already paid for its machines should keep producing as long as the revenue from one more unit exceeds the variable cost of that unit — even if average total cost is not yet covered. Shutting down avoids variable cost but gains nothing; the fixed cost is sunk. Conversely, a "profitable" line that would require a big marginal investment may be a bad next move if the marginal return is low. Thinking at the margin strips away what is irrelevant and focuses on the delta.
The discipline is asking "What changes if I do this?" and "What is the cost and benefit of the next unit?" It avoids the sunk-cost trap (we've already spent so much), the average trap (our overall margin is fine), and the all-or-nothing trap (we must be all in or all out). Used well, it improves pricing, capacity, hiring, and investment. Misused, it can ignore fixed-cost recovery or long-run effects that marginal analysis alone does not capture.
Marginal thinking is forward-looking. Historical cost, book value, and what you "should" get back are irrelevant to the next decision. The only question is whether the next unit of action adds more value than it costs. That unit can be one more widget, one more hire, one more dollar of investment, or one more hour of effort. Define the unit, estimate marginal cost and benefit, then decide.
Section 2
How to See It
Look for decisions that turn on the next unit: one more hire, one more shift, one more price change, one more store. The margin is visible when people ask "What do we gain or lose from this specific change?" rather than "Is the whole thing good?"
Business
You're seeing Thinking at the Margin when a company keeps a plant running at a loss because revenue per unit still exceeds variable cost per unit. Shutting down would save variable cost but would not recover fixed cost. The decision is made at the margin: one more unit of output adds more revenue than cost.
Technology
You're seeing Thinking at the Margin when a team ships an extra feature only if the incremental user value exceeds the incremental engineering and support cost. The question is not "Is the product good?" but "Does this next feature pay for itself at the margin?"
Investing
You're seeing Thinking at the Margin when an investor asks whether the next dollar of position size adds more expected return than risk, or whether the next hour of research is worth more than its opportunity cost. Portfolio and effort are adjusted at the margin, not in bulk.
Markets
You're seeing Thinking at the Margin when a central bank sets rates by weighing the incremental benefit of one more hike (lower inflation) against the incremental cost (higher unemployment, financial stress). Policy is a marginal decision, not a binary one.
Section 3
How to Use It
Decision filter
"Before committing, ask: what is the marginal cost and marginal benefit of this specific action? Ignore sunk cost and fixed cost for the decision at hand. If marginal benefit > marginal cost, do it. If not, don't. Then check that you are not missing long-run or strategic effects that don't show up in the next unit."
As a founder
Use marginal thinking for pricing (what does one more customer cost and yield?), capacity (what does one more unit of capacity cost and return?), and hiring (what does one more person add versus cost?). Avoid deciding by average margin or total profit when the real lever is the next increment. The mistake is doubling down because "we've already invested so much" — that is sunk cost. The next dollar and the next unit are what matter.
As an investor
Evaluate the next position, the next dollar of concentration, and the next hour of research at the margin. Marginal expected return versus marginal risk; marginal value of more due diligence versus opportunity cost. Portfolio construction is a sequence of marginal decisions. Avoid letting past positions or average performance dictate the next move.
As a decision-maker
Frame choices as marginal: what changes if we do this one thing? Strip out sunk and fixed elements. Compare incremental benefit to incremental cost. Use this for pricing, capacity, headcount, and capital allocation. When the decision has long-run or strategic consequences that the next unit does not reflect, add those explicitly — but keep the margin as the starting point.
Common misapplication: Letting sunk cost drive the next decision. "We've already spent X" is irrelevant to whether the next dollar is well spent. The margin is forward-looking.
Second misapplication: Ignoring that the margin can change with scale. The first unit and the millionth unit can have different marginal cost and benefit. Dynamic effects (learning, capacity constraints) may make the "next unit" depend on path. Use marginal thinking, but update the numbers as you go.
Buffett applies marginal thinking to capital allocation. Each dollar of retained earnings or new investment is evaluated at the margin: what is the next best use? He ignores sunk cost and historical cost when judging whether to hold or add to a position. The question is always incremental expected return versus incremental risk and opportunity cost.
Ed ThorpMathematician, author of Beat the Dealer and A Man for All Markets
Thorp used marginal analysis in blackjack (when does one more card justify one more bet?) and in investing (position sizing, Kelly criterion). The focus was always on the next decision: expected value and risk of the incremental action, not the total stake or past results. Thinking at the margin was the core of his edge.
Section 6
Visual Explanation
Thinking at the Margin — Optimal quantity is where marginal benefit equals marginal cost. Sunk and fixed cost stay out of the decision.
Section 7
Connected Models
Thinking at the Margin connects to cost-benefit analysis, sunk cost, and trade-offs. The models below either reinforce it (marginal cost/benefit, opportunity cost), create tension (sunk cost fallacy, incentives), or extend the logic (diminishing utility, trade-offs).
Reinforces
Marginal Cost/Benefit
Marginal cost and marginal benefit are the quantities that matter at the margin. Thinking at the margin is the discipline of using them: always compare the next unit's cost to the next unit's benefit. The two are the same framework — one is the concept, the other is the application.
Reinforces
Opportunity Cost
The cost of the next unit is what you give up to get it — the next best use of the same resource. Opportunity cost is the marginal cost of using a resource. Thinking at the margin forces you to make opportunity cost explicit.
Tension
Sunk Cost Fallacy
Sunk cost is the mistake of letting past cost influence the next decision. Thinking at the margin is the corrective: ignore sunk cost; only marginal cost and benefit matter. The tension is psychological — people want to "get their money's worth" — but the logic is clear.
Tension
Incentives
Incentives shape behaviour at the margin. When incentives are misaligned, people make the "rational" marginal choice from their perspective but the wrong one for the system. The tension is between marginal logic and the design of incentives so that marginal private benefit aligns with marginal social benefit.
Section 8
One Key Quote
"The marginal utility of a thing to a consumer is the benefit he gets from the last unit of it he has, or would get from adding one more unit."
— Alfred Marshall, Principles of Economics (1890)
Marshall put the margin at the centre of value and choice. The "last unit" and "adding one more" are the right units of analysis. Totals and averages follow from the margin; the decision rule is to adjust until the last unit is worth what it costs.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Ask "what changes?" for every decision. The margin is the change in outcome from the next unit of action. If you cannot state the marginal cost and marginal benefit, you are not yet thinking at the margin. Make the increment explicit.
Sunk cost is a trap. Past spend does not affect the optimal next move. The only question is whether the next dollar or the next unit is worth it. Train yourself and your team to ignore sunk cost when deciding forward.
Use it for pricing, capacity, and capital. Price where marginal revenue equals marginal cost (or the practical equivalent). Add capacity when the marginal return exceeds the marginal cost of capital. Allocate capital to the highest marginal expected return. The framework scales from a single product to a portfolio.
Watch for lumpy and long-run effects. Marginal analysis is cleanest when you can vary continuously and when the next unit does not change the game. When decisions are lumpy (e.g. one new plant) or when they have long-run strategic effects, add those to the marginal calculation rather than abandoning the framework.
Train the habit. Default to "what is the marginal cost and benefit?" in meetings and memos. When someone argues from total cost or past spend, reframe: "For the next unit, what do we gain and what do we give up?" The habit spreads and improves the quality of decisions across the organisation.
Section 10
Test Yourself
Is this mental model at work here?
Scenario 1
A plant is losing money on a full-cost basis. Management keeps it open because revenue per unit exceeds variable cost per unit.
Scenario 2
A team says we should finish the project because we've already spent $2m.
Scenario 3
An investor adds to a position only when the expected return on the next dollar exceeds the opportunity cost of that dollar.
Scenario 4
A team debates whether to add a fifth engineer. The discussion focuses on the output the fifth would add versus the salary and onboarding cost.
Summary. Decisions are made at the margin. The relevant question is the cost and benefit of the next unit, not totals or averages. Marginal benefit above marginal cost says do more; below says do less. Ignore sunk cost. Use the framework for pricing, capacity, hiring, and capital allocation. Watch for lumpy decisions and long-run effects that the next unit does not capture. Thinking at the margin is the discipline of asking "What changes if I do this?" and acting on the answer.
Kahneman documents how people often fail to think at the margin (sunk cost, framing) and how to correct for it. The psychology behind marginal discipline.
Accessible treatment of marginal cost, marginal benefit, and decision rules. Applications to production, consumption, and policy. Good for building the habit of marginal reasoning.
Leads-to
Diminishing [Utility](/mental-models/utility)
Marginal benefit often falls as quantity rises (diminishing marginal utility). That is why the marginal curve slopes down and why there is an optimal quantity. Diminishing utility is the reason the margin eventually bites; thinking at the margin is how you find where.
Leads-to
[Trade-offs](/mental-models/trade-offs)
Every marginal decision is a trade-off: a bit more of one thing, a bit less of another. Trade-offs are the content of marginal analysis. The model makes the trade-off explicit: what do we give up, what do we get, for the next unit?
