Skip to content
finalstate.aiAffordable Loss: Size the Bet by What You Can Lose
127Atlas

127The kitchen drawer

Affordable Loss: Size the Bet by What You Can Lose

Affordable loss limits a commitment to what the decision maker can survive losing under uncertainty.

An open kitchen drawer contains cash, a calendar and a tool.A cash envelope, available weekends and an owned tool form a proposed commitment. The drawer is an illustrative resource inventory; it cannot establish that liabilities outside it are bounded.Proposed commitment
Illustrative scene. The drawer contains proposed resources, not evidence that every exposure has been counted.

Size the commitment

Set the loss limit before choosing the commitment

The proposed cash, weekends and tool are lost while resources outside the commitment remain.A total-loss scenario crosses out the proposed commitment. A separate reserve remains, but its adequacy depends on the owner's obligations. No money amounts, resource ratios or survival finding are supplied.CommittedRemainingEnough?
Sarasvathy, Academy of Management Review 26(2), 2001, p. 252. The resource drawing is schematic; it assigns no value or probability.

In Sarasvathy's 2001 theory article, p. 252, affordable loss sets a limit in advance. It gives the decision maker a way to size an attempt when probabilities lack a reliable basis. Count cash, time, attention and any material reputational exposure. The owner must establish the full stake and what would remain after failure; the rule supplies neither a return estimate nor a reason to invest.

Reasoning under observation

The protocol recorded reasoning about an imaginary venture

A recorder and a marked transcript stand for the think-aloud study with 27 experts.The recording becomes a transcript that researchers classify. The selected sample and imaginary task constrain the evidence. Transcript marks are schematic and do not depict the disputed percentage summaries or outcomes of actual ventures.27 expertsRecordingCoded reasoning
Sarasvathy, Academy of Management Proceedings 2001, ENT:D4–D5; Effectuation.org, “The Venturing Experiment,” accessed 8 September 2026. The recorder and transcript are schematic; their marks show no frequency.

The archive and the Proceedings report give different descriptions of the research instrument. The latter lists five questions and tests its hypothesis using question 4. Their frequency summaries use different denominators, so no combined percentage is reported here. These records concern reasoning on a task; they do not measure venture success.

  • The programme archive describes 27 experts working through a 17-page problem with ten decisions.
  • The separate 2001 Proceedings paper analyses market-research responses in 235 coded meaning units.

Five principles

Affordable loss limits resources within a broader approach

  1. 01Bird in hand: begin with available means.
  2. 02Affordable loss: limit the resources committed.
  3. 03Crazy quilt: build with stakeholders who commit.
  4. 04Lemonade: work with unexpected developments.
  5. 05Pilot in the plane: act on what people can influence together.

These are the Society for Effectual Action's current teaching labels. The 2001 theory article sets out four underlying principles on p. 252. The named five-part model is a later teaching frame, and its affordable-loss principle includes time and other resources as well as money.

Available means, a commitment limit, joined partners, a diverted route and a steering wheel identify five teaching concepts.The commitment limit is the accent object. The surrounding means, partners, surprises and control concepts put affordable loss within the later five-principle teaching frame. They are not stages that every venture must pass.MeansCommitmentPartnersSurprisesControl
Effectuation.org, “The Five Principles of Effectuation,” accessed 8 September 2026. The five objects identify concepts, not a required sequence or a performance score.

An exposure record

An unanswered check leaves the commitment unresolved

An exposure record moves from open checks to provisional checks, then a guarantee invalidates the loss limit.Three checkboxes begin unanswered. The listed commitment and reserve support provisional cost and capacity checks. A newly found guarantee fails the boundary check and reopens the first two. The complete still shows the failed cap, unanswered revised checks and an untested upside.CostCapacityCapGuaranteeUpside untested
Illustrative exposure record. A tick means only that the stated check has supporting information in this example. The guarantee invalidates the proposed cap; no investment decision is approved.
  • Record every material cost and obligation.
  • Describe the capacity left after total failure.
  • Check for exposure beyond the proposed limit.

This editorial diagnostic applies the principle; it is not a validated Sarasvathy questionnaire. In the illustrative record, the listed resources fit the owner's stated reserve needs. A guarantee then reveals an additional obligation. The cost and survival checks must reopen, and the failed cap cannot support a commitment. Upside is still untested.

Given effect, given means

The starting point changes the choice being made

Several possible means converge on a fixed effect; one given set of means branches toward possible effects.The top relationship starts with the effect fixed and leaves means open for selection. The bottom starts with means fixed and leaves effects open. The two conditional logics can coexist; no path is marked as the winning choice.CausationMeansEffectEffectuationMeansEffects
Sarasvathy (2001), p. 245. Alternative routes and effects are schematic. Neither lane claims universal superiority.
  • Causation starts with a specified effect and selects means to produce it.
  • Effectuation starts with given means and selects among possible effects.

Sarasvathy defines this distinction on p. 245 and says the logics can coexist. Under Knightian uncertainty, starting from available means reduces dependence on a forecast. Where probabilities and values have a defensible basis, they still belong in the decision.

The wider theory remains contested

Arend, Sarooghi and Burkemper (2015) argued that effectuation needed stronger testing and theoretical development. Read, Sarasvathy, Dew and Wiltbank (2016) challenged the critique's assumptions and coverage of the research. This exchange supports treating the framework as contested. The exposure checks here are a practical application, with responsibility for their completeness left to the decision maker.

A limit on loss

A hidden obligation can exceed the stated loss limit

The same schematic payoff sits above two stated caps; a guarantee carries exposure through one cap.The left cap contains the illustrated commitment. On the right, a guarantee breaches the stated cap and extends exposure. The payoff sketches are not convexity tests, and the existence of a floor says nothing conclusive about curvature.Stated capCapGuaranteeExposure
Illustrative loss boundaries, not a measured payoff distribution. The guarantee adds exposure below the stated cap. No curvature result follows from either sketch.

The guarantee reaches beyond the proposed commitment. That defeats this strategy's loss limit, regardless of the payoff's shape. Mathematical convexity requires a separate chord test across a stated domain: a loss cap neither proves convexity nor is required for it. When Uncertainty Pays You keeps survivability, upside constraints and curvature separate.

Before the next commitment

A survivable loss leaves the opportunity to be judged

A loss record and an unresolved opportunity record remain separate, above an open decision marker.The close carries the exposure record to a separate appraisal of the opportunity. Neither record is marked approved. The owner still must judge value and establish any future right on its own terms.LossOpportunityDecision pending
The two records remain open for judgment. A completed loss assessment would not approve the opportunity or secure a future right.

Carry the exposure record back to the barbell's wild slice. Inspect costs, upside constraints and payoff shape separately. A bounded experiment also does not secure an option right: where a contract supplies that right, establish its enforceable terms, expiry and freedom to decline. Those are distinct requirements within optionality.

Read the transcript

01 · THE KITCHEN DRAWER

Late at a kitchen table, a woman is deciding whether to try a new venture. In this illustrative scene, she has too little evidence to put a useful number on the return. She pushes the laptop aside and opens a drawer. There is some cash, a calendar with free weekends, and a tool she owns outright. She begins by listing what she could commit. Losing that cash would hurt, and the weekends would be gone, but neither should put her ordinary obligations out of reach. That is only a proposed limit. Before she relies on it, she must find out whether the venture could cost her anything else.

02 · SIZE THE COMMITMENT

Affordable loss sets a limit before the commitment is made. Sarasvathy describes it on page two hundred fifty-two of her two thousand one theory article: the decision maker predetermines an affordable loss, then experiments within the available means. Apply total failure to the proposed commitment. The cash is spent, the weekends are used, and the tool is lost. What remains has to cover ordinary obligations and leave enough capacity for whatever comes next. The amount depends on the person and their circumstances. A small cash payment can still consume time or attention they cannot spare. Damage to reputation belongs in the assessment where it is a material exposure. This is a way to size exposure under uncertainty. Whether the attempt offers enough benefit remains a separate judgment.

03 · REASONING UNDER OBSERVATION

The principle grew out of research into how entrepreneurs reason. The programme archive describes twenty-seven selected experts working through a seventeen-page problem with ten decisions while thinking aloud. That archive describes experienced founders who had taken a company public. A separate two thousand one Proceedings paper reports an analysis of an imaginary venture called Venturing. It lists five questions and tests its hypothesis using the market-research question, divided into two hundred thirty-five units of meaning. These are different descriptions of the research instrument, and their published percentage summaries use different denominators. A single frequency would conceal that problem. The limited conclusion is about reasoning observed on the task. Neither document makes those percentages a venture success rate. The Academy of Management Review article, published in the same year, develops the theory.

04 · FIVE PRINCIPLES

Affordable loss sits within a broader approach to making a venture. The Society for Effectual Action now teaches five principles. Bird in hand begins with who you are, what you know, and whom you know. Affordable loss limits the resources committed. Crazy quilt concerns stakeholders who join through commitments of their own. Lemonade asks how an unexpected development could be used. Pilot in the plane concerns what people can influence and create together. These names are the later teaching frame. The two thousand one theory article sets out four underlying principles on page two hundred fifty-two. The commitment limit applies to more than money: our woman at the kitchen table must account for the weekends and her attention too.

05 · AN EXPOSURE RECORD

The kitchen-table inventory can become an exposure record. This is an editorial diagnostic derived from affordable loss, not a validated questionnaire. Start with the full cost of the attempt. In this example, she has listed the cash, weekends and tool, and compared their loss with the resources needed for her ordinary obligations. The listed commitment passes those checks on the information recorded. The boundary check is still pending. Then she finds a guarantee in the proposed contract. It makes her responsible for a supplier's unpaid bill if the venture fails, beyond the resources she had listed. The guarantee was missing from her limit. That fails the cap check and reopens the cost and survival checks. She needs a revised exposure record, or different terms, before relying on the limit. Upside is still untested, and even a completed record would leave the investment decision to her.

06 · GIVEN EFFECT, GIVEN MEANS

The formal distinction concerns what is given at the start. Causation begins with a specified effect and selects means to produce it. Effectuation begins with given means and selects among possible effects. For the woman at the table, a fixed product plan would send her looking for the resources it requires. Starting from her available resources leaves the product itself open to revision. Sarasvathy makes both definitions on page two hundred forty-five and says the logics can coexist and intertwine. A forecast with defensible inputs can inform a decision. Affordable loss becomes useful when the owner cannot support the return estimate on which the commitment would otherwise depend.

07 · Advertisement · Twain GTM Agents

A good outbound message should know why it is being sent. Twain GTM agents research accounts, qualify contacts, and turn live signals into relevant outreach. They work with your CRM, personas, and GTM stack, so the message is not just personalized. It is backed by a specific reason to start the conversation.

08 · A CONTESTED FRAMEWORK

The wider theory remains contested. In twenty fifteen, Richard Arend, Hessam Sarooghi, and Andrew Burkemper argued that effectuation needed stronger empirical testing and theoretical development. Stuart Read, Sarasvathy, Nicholas Dew, and Robert Wiltbank replied in twenty sixteen. They challenged the assumptions used to assess the theory and argued that the critique had omitted relevant research. The exchange leaves a dispute about how the theory should be judged and developed. The narrower use here is to make a proposed commitment and its consequences explicit. A named framework cannot establish that the owner has found every liability or kept enough capacity in reserve.

09 · A LIMIT ON LOSS

A guarantee can carry the loss beyond the amount the owner intended to commit. The stated cap no longer contains the exposure. That is a failure of this loss-limited strategy. It says nothing by itself about mathematical convexity. Convexity concerns whether a payoff lies on or below every chord across the domain being tested. A capped loss neither proves that property nor is required for it. Keep the questions separate: what can be lost, what return is available, and what shape describes the payoff? A capacity limit or a partner's veto can constrain the upside without adding to the loss. Those constraints need their own assessment even after the exposure record is complete.

10 · BEFORE THE NEXT COMMITMENT

Back at the kitchen table, she now knows why the drawer was only a starting point. The proposed commitment must include obligations outside it, and total failure must leave enough capacity for her continuing responsibilities. Once that assessment is complete, she can judge the opportunity: its costs, constraints on the upside, and payoff shape over the relevant range. A bounded experiment does not itself secure the right to make a later investment. If that right comes from a contract, she must establish the enforceable terms, the expiry and her freedom to decline. Affordable loss supplies a limit on what she commits. She still has to decide whether this opportunity deserves any of it.

01 / 10 · THE KITCHEN DRAWER0:00 / 7:53