Concept

Two-Way Door Decisions

conceptdecision-makingleadershipbusinessregulation
Teach me this →

Two-Way Door Decisions

Jeff Bezos‘s framework for calibrating how much analysis a decision deserves and who should make it, built on a single test: can you walk back through the door if you chose wrong?

A two-way door decision is reversible — you pick a path, discover it was wrong, and go back to pick another one at acceptable cost. A one-way door decision is not: once you walk through, you cannot return, or returning is so costly and slow that it amounts to the same thing. Bezos traces most organisational slowness to a failure to tell these apart: companies default to a single, heavyweight decision-making process and apply it to everything, including the large majority of decisions that are actually two-way doors. ‘Most of what slows things down is taking too long to make decisions at all scale levels,’ he argues — the fix is matching the weight of the process to the reversibility of the decision, not running every decision through the same machinery.

The governance rule

Two-way door decisions should be made quickly, by single individuals or small teams deep in the organisation, on the understanding that a wrong call can be corrected later at low cost. One-way door decisions — the rare ones that are genuinely hard or impossible to reverse — should be elevated to the most senior people available, made deliberately, and deliberately slowed down. As CEO of Amazon, Bezos describes his own role on these as ‘chief slow-down officer’: when a one-way door decision arrived at his desk, his job was to insist on more analysis — ‘I can think of three more ways to analyze that’ — precisely because the decision could not cheaply be undone.

The test that sorts a decision into one category or the other is not its apparent size but its reversibility. Bezos gives two examples from different domains: the decision to launch Amazon Prime was consequential and expensive to unwind, so it merited one-way-door scrutiny even though it was, in form, a product launch. Choosing the propellants for a new rocket — liquefied natural gas for the New Glenn booster stage, liquid hydrogen for the upper stage — is what he calls ‘quick-drying cement’: technically it can be changed, but changing it after the fact is such a large setback that it functions as a one-way door.

Why the distinction matters for velocity

Bezos frames this explicitly as a solution to organisational drag at scale. Amazon, with over a million employees, stays fast because the culture supports routing two-way-door decisions to the people closest to the problem instead of up a chain of approvals. The distinction only works if the organisation can actually tell the two kinds of decision apart in practice — a skill Bezos treats as itself a core leadership competency, not a formality to apply after the fact.

Extension: the one-way door in regulation

Blake Scholl applies the same reversibility test outside the corporation, to public regulation, and supplies a mechanism Bezos’s corporate framing does not: why a one-way door, once installed, stays shut. Bezos’s one-way doors persist because reversal is technically or financially costly — quick-drying cement. Scholl’s regulatory one-way doors persist for a different reason entirely: reversal is politically costly to the individual who attempts it, regardless of the underlying merits. ‘Regulations, particularly anything that’s got a tie to safety, tends to be a one-way door,’ he argues — the official who removes a safety-branded rule bears the full blame if anything later goes wrong, while the official who leaves a useless rule in place bears no comparable cost, however great the rule’s ongoing waste. Bezos’s chief-slow-down-officer role works because a single accountable executive can weigh the two paths and choose to invest more analysis; Scholl’s diagnosis is that no equivalent single decision-maker exists in public regulation, so the asymmetric incentive runs unchecked and rules accumulate rather than get pruned. He offers airport security screening — persisting two decades after its stated rationale, in his account, stopped applying — as the clearest case, and extends the same logic to Congressionally-scattered defence-manufacturing supply chains, where votes rather than reversibility-of-outcome decide the shape of the system. See Blake Scholl on Supersonic Flight and Fixing Broken Infrastructure.

Where mainstream views differ

The framework is now widely cited in management writing (it appears, unattributed to any page in this wiki until now, in discussions of platform-versus-product decisions and career moves alike). The main line of critique is that reversibility is rarely binary: many decisions sit on a spectrum of cost-and-speed-of-reversal rather than splitting cleanly into two doors, and labelling a decision ‘one-way’ can become a way for senior leaders to justify retaining control they would otherwise have to delegate. Bezos’s own examples partly concede the point — Amazon Prime’s reversibility is, in his words, ‘unclear’ — suggesting the test is a heuristic for triage rather than a strict binary classification.

In the wiki