Stop Searching for the Perfect Decision. Nobel Prize Research Says "Good Enough" Wins.

The best-documented satisficing rule in modern business sits in a shareholder letter. In his 2015 letter, Jeff Bezos split decisions into two kinds: Type 1 — consequential, irreversible or nearly so, one-way doors, to be made methodically, carefully, slowly — and Type 2 — changeable, reversible, two-way doors, where a suboptimal call can simply be walked back. His diagnosis: as organizations grow, they run the heavyweight Type 1 process on everything, and the result is slowness and diminished invention. The following year's letter added the number: most decisions should be made with somewhere around 70% of the information you wish you had.

Bezos was writing operating advice, but he was restating a Nobel-winning result. Herbert Simon proved in the 1950s that exhaustive search for the best option is not what rational decision-making looks like — not for people, not for organizations, not even in principle. He named the actual strategy satisficing: satisfy plus suffice. Define what good enough means, search until you find it, stop.

Now look at the comparison spreadsheet currently open somewhere in your organization — fourteen vendors, forty criteria columns, a decision meeting that keeps rescheduling. This piece is about why that spreadsheet is usually the expensive mistake, what the evidence says it costs the people who build it, and how to satisfice deliberately — including the cases where you genuinely should maximize. The theory behind it all has its own companion piece: Simon's bounded rationality.

Doing Better, Feeling Worse: What Maximizing Actually Buys

The classic test tracked university students through their final-year job search. Sheena Iyengar, Rachael Wells and Barry Schwartz measured who approached the search as maximizers — determined to find the best possible job — and who satisficed against their own criteria. The maximizers won on paper: starting salaries about 20% higher. And they lost everywhere else: less satisfied with the jobs they took, more pessimism, stress and regret throughout the process (Psychological Science, 2006 — the paper's title says it: Doing Better but Feeling Worse).

The broader maximizing literature, running from Schwartz's original 2002 studies through recent reviews, repeats the pattern: maximizing tendencies correlate with more regret, more decision difficulty, more depression and less life satisfaction — with regret doing much of the damage. A 20% salary bump is real money and may be worth it once, at a career hinge. As a default operating mode for a team, maximizing buys marginal object-level gains at a compounding subjective and organizational cost: slower decisions, exhausted deciders, and choices everyone second-guesses because the process promised best and best is unfalsifiable.

The Math Was Never on the Spreadsheet's Side

Even if search were free of regret, it isn't free of time — and the mathematics of when to stop searching has an exact answer for an idealized case. In the classic secretary problem (rank candidates seen one at a time, no going back), the provably optimal strategy is to observe the first 37% of options without committing, then take the first one better than everything seen so far. Not 100%. Not most. Thirty-seven percent — and that is the optimal policy, in a setting engineered to favor thorough search.

Real decisions are messier than the theorem, but its lesson transfers: rational search examines a fraction of the option space and commits. Three forces guarantee it. Search costs are real — every evaluation hour has an opportunity cost. Returns diminish — the difference between the 90th and 95th percentile option is small and shrinking. And delay compounds — the market a decision was meant to serve moves while the decision is being perfected. A good decision executed for two months usually beats a slightly better one that arrives two months late.

The Satisficing Framework: Five Steps

Satisficing done deliberately is a process, and the discipline is front-loaded:

When to Maximize Instead: The One-Way Doors

Satisficing is a default, not a dogma, and the exceptions are exactly Bezos' Type 1 decisions. Maximize when the decision is irreversible or nearly so (an acquisition, a platform migration, a co-founder); when the option space is small enough to actually enumerate; when nothing forces speed; or when tiny quality differences compound for years. Those decisions deserve the full apparatus — wide search, structured debate, devil's advocates, time.

The failure mode is almost never maximizing a two-way door too little. It is running the one-way-door process on two-way doors — fourteen-vendor spreadsheets for a tool you could swap out next quarter. The skill is triage: classify the door first, then pick the process. Product teams will recognize satisficing under its other names — the MVP, the iteration, Reid Hoffman's famous advice that a first version which doesn't embarrass you shipped too late. Agile is satisficing applied to scope: ship good enough, learn, adjust — because the market lesson you buy by shipping is worth more than the polish you'd buy by waiting.

But What If We Miss Something Better?

You will. That is not a flaw in satisficing — it is a fact about option spaces, and it holds for maximizers too. There is always another vendor, another candidate, another framework; the exhaustive search that would rule them all out does not terminate. Maximizing doesn't close the gap, it only moves the moment of stopping later and prices it higher.

And the fear points the wrong way, because the regret data runs against the maximizers, not the satisficers. The people most determined not to miss something better are the ones the studies find least satisfied with what they chose. Well-set aspiration levels are the honest answer to the fear: if something genuinely important would be missed, it belongs in the criteria — and then satisficing protects it better than an open-ended search does, because the criteria are written down and checked instead of floating in the room.

The residual risk is real but different: aspiration levels can be set badly. Too low, and you accept the mediocre; too high, and nothing ever passes — which is decision paralysis wearing a process costume. Calibration comes from the record: when you document criteria and outcomes, you find out which bars were wrong and move them — that is what the decision log is for.

Where Argumentree Fits

The hardest part of satisficing is not understanding it — it is trusting the stop. Teams keep searching because nobody can see that the bar was genuinely cleared. Argumentree makes the stop trustworthy: the criteria live in a shared structure everyone sees before options arrive; the arguments for and against each option are laid out and rated instead of re-argued from memory; and the record — criteria, evaluations, reasoning — persists, so the decision doesn't get re-opened every time someone new hears about a shinier alternative.

Visible criteria, weighed arguments, a durable record: that is the environment in which good enough is a defensible standard instead of a hopeful shrug.

Start With One Decision

Pick one decision your team is currently maximizing. Classify the door — if it's reversible, write the criteria, set the budget, take the first pass, and log it. The discomfort you feel at stopping is not the feeling of doing something wrong; it is the feeling of doing something different, with a Nobel Prize, a stopping theorem and a stack of regret studies behind it.

Simon's standard was never mediocrity. It was reliability: the organization that decides well consistently beats the one that decides perfectly occasionally, because the second one doesn't exist.