---
name: Satisficing
status: draft
territory: decision-making-under-uncertainty
host_mode: constraint-mapping
also_loadable_in: []
msi_wired: false
sources:
  - title: Simon, Herbert A. (1956), "Rational Choice and the Structure of the Environment," Psychological Review 63(2):129-138
    url: https://doi.org/10.1037/h0042769
  - title: Simon, Herbert A. (1955), "A Behavioral Model of Rational Choice," Quarterly Journal of Economics 69(1):99-118
    url: https://doi.org/10.2307/1884852
---

# Satisficing

## Why it matters

Trying to find the *best* option is often the worst possible strategy — you can't examine them all, the search costs more than the prize, and the smart move is to decide in advance what "good enough" means and grab the first option that clears it.

For example: a team needs to pick a logging library and there are a dozen credible ones. They could benchmark all twelve, read every open bug, and build a proof-of-concept for each — two weeks of engineering gone, to find a "best" whose edge over the runner-up is a rounding error. Instead they write the bar down first: structured output, actively maintained, handles ten thousand events a second, has real docs. They check libraries one at a time, and the third one clears all four. Ship it. Nothing about the next nine would have paid for the time spent looking — and the choice that would have looked smartest in a textbook is the one that wastes the most of what's actually scarce.

- **What it reveals.** That a sound decision can be made by *stopping* — set a "good enough" bar, search until one option clears it, take that one — and that the cost of searching, not just the quality of options, is part of the choice.
- **How it changes the read.** You stop asking *"which option is best?"* and start asking *"what threshold must an option clear, and is the search past that point worth its cost?"* — comparing options against a bar, not against each other to the last decimal.
- **When to foreground it.** A large or expensive-to-survey option space; small differences between the good options; a reversible choice; or a decision stalled in analysis paralysis with nothing being chosen at all.
- **What you'd miss without it.** That past the threshold, more search usually buys nothing — and that "optimal" choices often cost more in time, attention, and second-guessing than the better-on-paper option is worth.
- **Where it misleads.** Set the bar too high and you never stop; set it too low and you ship junk. And it is a rule about the *cost of searching*, not a license for low standards — misread as "good enough is fine," it excuses settling; used on a high-stakes, irreversible, cheap-to-search choice, it under-searches a decision that genuinely warranted optimizing.

## How to invoke it in Ora

You have several real options on the table and you want them laid out side by side — each one's tradeoffs and where it's truly constrained — without Ora picking for you. Satisficing rides along inside that comparison: it's one of the thinking tools the mode always has on hand, and it supplies the discipline of judging each option against a "good enough" bar instead of chasing the maximum on every dimension.

Describe the options and the dimensions you care about, and ask:

> "Compare the alternatives for our cloud migration — AWS-only, multi-cloud, hybrid. Map the tradeoffs across cost, lock-in, and team skill."

Ora maps each alternative on the same four slots — what would make it succeed, what would make it fail, what it uniquely gains, what it forfeits — and then compares them across your dimensions. Satisficing is what keeps that comparison honest: for each dimension it asks whether an option *clears the threshold* rather than whether it tops the field, so two options that both clear "cost decreases over three years" are treated as tied on cost, and the comparison spends its attention where the options actually differ in a way that matters. It also sharpens the *no-lose elements* — the moves worth making whichever option you pick.

One thing to know: the words *compare alternatives*, *map the tradeoffs*, *what are the pros and cons of each option* are what route you here, to the host analysis — not the phrase "satisficing" on its own. You don't summon this model directly; you ask for the alternatives to be mapped, and the model is always available inside it. Give it real, distinct options and the dimensions that matter; the comparison is only as good as the bar you can name for each dimension.

Say what "good enough" looks like on each dimension if you can — a rent ceiling, a latency budget, a year-over-year cost line. The single most useful thing you can supply is the threshold, because that is what turns a vague "which is better?" into a clean "which clear the bar, and where past it does the difference stop mattering?"

One thing Ora won't do here: pick the winner. The mode maps the terrain rather than deciding — its recommendation slot stays empty unless you explicitly ask for a final choice — and satisficing serves the *map*: it tells you which options clear which bars and where the differences are real, leaving the stop-here judgment to you.

## How it works

A family wants to move, and they open the listings for a big city. There are thousands. They could, in principle, tour every one — but the listings change daily, a good apartment is gone within the week, and a lifetime would not be enough to walk through them all and rank them. Even the attempt is self-defeating: by the time you finished surveying everything, the place you'd have chosen on day one would have been signed by someone else.

So what do they actually do — and what does anyone sensible do? They decide, *before* they start looking, what would be acceptable. Under two thousand a month. Two bedrooms. Walkable to the train. Nothing on the ground floor. Then they go and look, and they sign the first apartment that clears every one of those lines — knowing full well that somewhere in the thousands they didn't see, a marginally nicer place at the same price almost certainly exists. They will never find it, and they were never going to. This is not laziness. It is the only rational thing to do once you take seriously that *searching has a cost* — in time, in attention, in apartments rented out from under you while you keep looking. The whole point of the search was a place to live, not the single best place in the city; that prize was never available, and chasing it would have cost them the apartment they could actually get.

Herbert Simon — an economist and psychologist who built his career on watching how people decide rather than how a theory said they should — gave this move its name. To **satisfice** (a word he coined by blending *satisfy* and *suffice*) is to set a bar of acceptability — an **aspiration level** — and take the first option that clears it. The alternative, **optimizing**, means examining every option and scoring each against every criterion to find the single best. Optimizing is what the economics textbooks of his day assumed every decision-maker did. Simon's objection was blunt: for any choice with real complexity, no actual mind — and no organization, and no computer of his era or ours — can do it. The number of options times the number of criteria explodes past anything a finite searcher can survey. So the question is not *how do we optimize?* It is *where do we stop?* And the answer real people reach, sensibly, is: at the first thing good enough.

The reveal underneath the apartment hunt is this: good deciders don't maximize — they satisfice. And the genuine skill is not in the searching at all; it is in **calibrating the aspiration level**. Set the bar too high and nothing ever clears it: you tour forever, paralyzed, and the apartments keep slipping away. Set it too low and you sign the first dump you see and regret it for a year. The whole craft is picking a bar that's demanding enough to protect you and loose enough to actually be met — and then *committing* to it, because a bar you quietly raise every time a candidate appears is no bar at all. Simon also noticed something that stings: the people who insist on finding the best — the maximizers — often end up *less* satisfied than the satisficers, even when they objectively choose better, because they can't stop measuring their choice against the ones they didn't take. The satisficer, who decided in advance what good enough meant and took it, simply moves into the apartment.

This is the decision strategy that follows directly from **bounded rationality** — Simon's broader account, explained on its own page, of how finite minds with finite information actually choose. Bounded rationality is the diagnosis: the mind is small next to the problems the world hands it. Satisficing is the prescription that falls out of it: set a bar, search to it, stop. Simon won the Nobel Prize in Economics in 1978 for the larger body of work, and the idea has only hardened with time, because the constraint it rests on never lifts — there will always be more options than anyone can weigh, and the realistic measure of a good decision is never "did they find the best?" but "did they set the right bar, and did they have the discipline to stop when something met it?"

## Framework & implementation

*This section uses Ora's own terms for the parts of an analysis, so that if you open the actual mode and lens files they line up. Each is glossed in plain language on first use.*

### Pipeline execution

Satisficing is one of the **always-loaded mental models** of the Constraint Mapping analysis — a thinking tool the mode keeps on hand for every run, listed in the mode's **`ANALYTICAL PERSPECTIVES`** block under "always loaded," alongside bayesian-reasoning, trade-offs, bottlenecks, and leverage. Like all the mode's mental models, it supplies no section of its own: the output skeleton is the mode's, and satisficing informs the *comparison* from inside it. Constraint Mapping is a *compare-the-alternatives* analysis — it lays several viable options side by side and **maps the terrain rather than deciding** (its conditional recommendation stays empty unless the user explicitly asks for a choice). Satisficing is the lens that governs *how each option is judged against a dimension*: by whether it clears a "good enough" bar — an **aspiration level** — rather than by whether it maximizes. The mode runs at **Gear 4**, Ora's most thorough setting — a **Depth analyst** and a **Breadth analyst** map the options independently, each critiques the other's reading, both revise under that critique, a flagged factual claim is web-checked, and a consolidator and formatter assemble the result. The model threads through those stages like this.

**Detection.** The model becomes load-bearing on the cases in its **Detection Signals** — the option space is large and the differences between the *good* options are small; the cost of evaluating one more option exceeds its likely improvement; the decision is reversible, so optimizing upfront is waste; or analysis paralysis has set in and nothing is being chosen at all. Inside a constraint-mapping run these show up as dimensions where every viable alternative comfortably clears the relevant bar, and squeezing the last unit out of one option buys nothing the decision actually needs.

**The Depth and Breadth analysts.** Two models map the alternatives in parallel; neither sees the other's work. As each fills in the mode's **per-alternative analysis** — the symmetric four-slot reading of *success conditions*, *failure conditions*, *uniquely gained*, and *forfeited* for every option — satisficing runs the lens's **Application Steps** against each dimension. It asks the steps in order: was the minimum acceptable criterion — the threshold — defined *before* the options were scored? Does each option clear it? Where two or more options clear the same bar, are they being treated as equivalent on that dimension rather than separated by a difference past the threshold that doesn't matter? The **Depth analyst** presses one dimension hard — is this a case where the threshold is the right target, or one of the high-stakes, irreversible exceptions where real optimization is warranted? The **Breadth analyst** checks the bar across all the dimensions and all the options at once, so a threshold isn't applied strictly to one alternative and loosely to another. The model never overrides the four-slot structure — it supplies the discipline for reading it.

**What it contributes to the comparison.** Satisficing does its most visible work in the mode's **cross-alternative comparison** and its **cross-alternative differentiating factors**. By treating each dimension as a threshold rather than a maximization contest, it collapses the dimensions where every option is good enough — and *concentrates the comparison on the dimensions where the options genuinely diverge*, which is exactly where the differentiating factors live. It pairs naturally with **trade-offs** (the always-loaded model for what each option forfeits to gain something else): satisficing says where on a dimension "enough" is, and trade-offs says what clearing a higher bar there would cost elsewhere. It also sharpens the **no-lose elements** — moves valuable under every alternative — because an action that lifts every option comfortably past a shared threshold is a no-lose move, whereas one that only pushes a single option from "good enough" to "slightly better" usually is not.

**The optimize-versus-satisfice judgment.** The lens carries an explicit fork, and Gear 4's two analysts argue it out: most decisions warrant satisficing, but some warrant optimizing — specifically the ones that are high-stakes, irreversible, and cheap to search, where the variance between options is genuinely large. The analysis is expected to *classify the decision's stakes before choosing the strategy*, not default to "good enough" out of fatigue. On a reversible, low-variance choice it applies the threshold and stops; on a one-way, high-consequence door it flags that exhaustive comparison is the right move and that satisficing here would under-search.

**Cross-adversarial evaluation.** Each analyst's reading is handed to the *other* to critique, and the model's signature failures — keyed to its **Common Failure Modes** — are caught here. The chief one is **threshold drift**: a bar that quietly gets more demanding as candidates appear, so the first acceptable option is disqualified after the fact — the tell is threshold language tightening across the draft. The opposite framing error is **inappropriate satisficing**: applying the "good enough" rule to a high-stakes, irreversible decision where optimization was warranted — the tell is a satisficing call on a one-way door with large variance between options. The evaluator also catches **post-decision rumination** smuggled into the analysis as a reason to keep searching past a met threshold, and the deeper confusion the lens warns against most — treating satisficing as *laziness* or as a license for a low bar, when it is a claim about the **cost of search**, not the **level of the standard**.

**Revision and claim-check.** The reviser locks each threshold to where it stood before the options were scored, restores any dimension the draft turned back into a maximization contest, and makes the stakes classification explicit where the draft satisficed (or optimized) without saying why the decision warranted it. Where a reading rests on a checkable factual claim about an option — a benchmark number, a price, a maintenance status — that claim is flagged and sent to a web-search tool before the revised draft proceeds.

**Consolidation and output.** The model contributes no section of its own; its work shows up *inside* the mode's set sections, organized by the consolidator and placed by the formatter. The thresholds it enforces render as the testable propositions in each option's **success conditions** and **failure conditions**. Its dimension-by-dimension "clears the bar / doesn't" reading shapes the **cross-alternative comparison** table and is the engine behind the **cross-alternative differentiating factors**, which surface precisely the dimensions where the options stop being interchangeable. Its judgment about where "enough" is feeds the **no-lose elements**. The **analytical-depth-symmetry note** — the mode's check that no option was analyzed more deeply than the others — is exactly the guard satisficing needs against a threshold applied strictly to one alternative and loosely to another. And the **conditional recommendation** stays empty by default: satisficing maps which options clear which bars; it does not, on its own, pick the one to take.

**What the analysis will not assert.** It will not raise a threshold mid-comparison to disqualify an option that already cleared it; it will not treat a difference *past* a met threshold as a deciding factor when the decision doesn't need it; and it will not let "good enough" stand in for a careless or under-set bar — the rule licenses *stopping the search*, never *lowering the standard*.

### Origin and evidence

The model is Herbert A. Simon's. He set it out in two papers in the mid-1950s. "A Behavioral Model of Rational Choice" (1955), in the *Quarterly Journal of Economics*, replaces the optimizing agent of classical economics with a decision-maker who has limited information and limited computing power and therefore searches sequentially against an **aspiration level** — a "good enough" bar — rather than evaluating all options at once. "Rational Choice and the Structure of the Environment" (1956), in *Psychological Review*, develops the companion claim that this works because the structure of real environments lets a finite searcher do well without surveying everything — the aspiration level need only be matched to what the environment actually offers. He gave the strategy book-length treatment, and named the dynamics of how aspiration levels adjust to feedback, in *Models of Man: Social and Rational* (1957), and restated the position for an economics audience in his 1978 Nobel Memorial Prize lecture, "Rational Decision Making in Business Organizations" — the prize itself awarded for this body of work on decision-making in organizations. Satisficing is the operational core of his broader theory of **bounded rationality** (treated on its own page): bounded rationality is the diagnosis that the chooser is finite; satisficing is the tractable decision rule that follows — set a bar, search until the first option clears every bar, then stop. A later strand the lens names as adjacent is the *maximizer–satisficer* research in psychology, which finds that people disposed to seek the best option often report lower satisfaction with their choices than those who stop at "good enough," because awareness of the unchosen alternatives breeds regret. Inside a constraint-mapping comparison, Simon's claim does a precise job: it converts "which option is best?" into "which options clear the bar, and where past it does the difference stop mattering?"

### Applications and common uses

Satisficing is a working model anywhere a choice involves more options than can be fully surveyed and a defensible "good enough" bar can be named — and, inside a constraint-mapping comparison, anywhere options should be judged against a threshold rather than ranked to the last decimal.

- **Comparing alternatives against a bar.** Its native job in this host: when several viable options are mapped side by side, satisficing sets the "good enough" line on each dimension so that options clearing the same bar are treated as tied there, and the comparison's attention goes to the dimensions where they actually differ — the difference that drives the choice.
- **Vendor, tool, and library selection.** Defining required capabilities up front, evaluating candidates one at a time, and committing to the first that meets every requirement — instead of benchmarking the entire field to separate options whose remaining differences won't affect the outcome.
- **Hiring and search.** Setting the must-have bar for a role and extending the offer to the first candidate who clears it, rather than holding out for a mythical perfect hire while strong candidates take other jobs and the role stays open.
- **High-volume operational decisions.** Reversible, repeated, low-variance choices — which of several adequate suppliers, routes, or configurations — where the cost of optimizing each one would dwarf the spread between the good options.
- **Knowing when *not* to satisfice.** The model's discipline cuts both ways: it justifies optimizing on the decisions that warrant it — high-stakes, irreversible, cheap-to-search choices with large variance between options — by making the contrast with the satisficing default explicit, so the rare optimization is a deliberate call rather than an accident.

In every case the contribution is the same: a question of "which option is best?" is re-described as "which options clear the bar, where do the differences past it stop mattering, and is this a decision that warranted stopping at good enough?" — and the search is sized to the stakes rather than run to exhaustion by default.

### Failure modes and when not to use it

The model's characteristic ways of going wrong are catalogued in its **Common Failure Modes**:

- **Threshold drift.** The "good enough" bar is quietly raised mid-search to disqualify the first option that cleared it. The tell is threshold language getting more demanding as candidates appear. The correction: lock the threshold before the search begins; a bar you keep moving is not a bar, and the move usually masks a maximizer's reluctance to stop.
- **Inappropriate satisficing.** The rule is applied to a high-stakes, irreversible decision where real optimization was warranted. The tell is a satisficing call on a one-way door with large variance between options, producing regret because the spread between options genuinely mattered. The correction: classify the decision's stakes *before* choosing the strategy — satisficing is the default, not the universal answer.
- **Post-decision rumination.** The choice is second-guessed by going back to evaluate the options that weren't taken. The tell is regret rising *after* a threshold was met and a decision made. The correction: commit to not re-evaluating; the satisficing advantage — and most of its psychological payoff — disappears the moment you start measuring the chosen option against the ones you passed up.
- **Satisficing-as-laziness.** "Good enough" is treated as an excuse to set the bar low, collapsing a claim about the *cost of search* into a claim about the *level of the standard*. The tell is a threshold set well below what's readily available, defended as "we satisficed." The correction: the bar's height is set by the stakes; satisficing only licenses *stopping the search once that bar is cleared*, never lowering it.

**When not to reach for it.** When the option space is small, closed, and cheap to survey in full — a handful of well-specified choices where exhaustive evaluation costs almost nothing — satisficing adds little, because optimizing was actually feasible and the threshold frame just hides a comparison you could have run completely. When the decision is high-stakes, irreversible, and the variance between options is large, the model's *own* discipline says optimize, not satisfice — reaching for "good enough" there inverts it. And when the real task is to *pick the winner* among the mapped alternatives rather than to judge each against a bar, satisficing sharpens the comparison but does not make the choice — that is the user's call, or a stance-bearing mode's, not the constraint map's.

## Related

- **Constraint Mapping** — the analysis this model is loaded inside; it maps several alternatives' tradeoffs and binding constraints side by side, mapping the terrain rather than deciding, and satisficing supplies the "good enough" bar each option is judged against.
- **Bounded Rationality** — the parent theory: Simon's account of why finite minds can't optimize. Satisficing is the decision strategy that follows from it — set a bar, search to it, stop.
- **Trade-offs** — the comparison partner also always loaded in the mode: satisficing says where "enough" is on a dimension; trade-offs says what clearing a higher bar there would cost elsewhere.
- **Bottlenecks** — the sibling constraint-finding lens: where bottlenecks locates the one stage that caps an option, satisficing decides whether "good enough" on that constrained dimension is the right target rather than squeezing the last unit of throughput.

## Sources

- [Simon, Herbert A. (1956), "Rational Choice and the Structure of the Environment," Psychological Review 63(2):129-138](https://doi.org/10.1037/h0042769)
- [Simon, Herbert A. (1955), "A Behavioral Model of Rational Choice," Quarterly Journal of Economics 69(1):99-118](https://doi.org/10.2307/1884852)