Tool/Mental Model/No. 1057
Two-Way vs. One-Way Doors
Two-way vs. one-way doors is a management heuristic popularized by Jeff Bezos that distinguishes reversible decisions from hard-to-reverse ones. Two-way doors allow a return at an acceptable cost; one-way doors do not, so the distinction links the pace of decisions to their recoverability.
Also called One-Way and Two-Way Door Decisions · Reversible vs. Irreversible Decisions
- Evidence
- Useful, modest evidence
- Read
- 6 min
- Links
- 12 connections
01You've seen this when…
- in life
You want to move to a new city. A month-long rental lets you try the neighborhood; buying an apartment commits your savings before you know the commute.
- at work
A team puts a minor website change through the same approval process as a three-year supplier contract. Both wait on the director’s desk.
- out in the world
A library considers opening on Sundays. Trying it for six weeks is one decision; signing a long-term staffing contract to support it is another.
02The idea
A decision that takes minutes to reverse shouldn’t need the same process as one that locks you in for years. Yet teams often demand the same level of certainty for both decisions and send them through the same meetings and approvals.
The door metaphor separates them. A two-way door lets you return at an acceptable cost after acting and seeing what happens. A one-way door leaves consequences you cannot readily undo. Selling a business or demolishing a building makes reversal difficult or impossible, and you cannot readily take back private information once you disclose it.
Jeff Bezos called these Type 2 and Type 1 decisions, respectively. His advice was to let capable individuals or small teams make reversible decisions quickly, while giving hard-to-reverse decisions more deliberation and consultation.
The distinction turns on recoverability, not on how large or risky a decision is. A large experiment can be reversible if its exposure is bounded, while a tiny disclosure can be irreversible once someone has read it.
Treat the categories as a spectrum. Most doors open back only at a price. The useful question is whether you can afford that price, and whether you can get back before lasting harm occurs.
03How to use it
- Define the actual commitment. Separate exploring an option from committing to it. Interviewing a supplier differs from running a pilot, which in turn differs from signing an exclusive contract, even under the same project name.
- Price the return trip. List the money, time, disruption and obligations involved in reversing course. Include your own costs and the effects on customers and other people. A cancellation clause may restore your freedom without restoring their trust.
- Check the recovery window. Work out how soon trouble would become visible and how long reversal would take. Being able to stop next month is little help if serious damage can happen tomorrow.
- Build a smaller door. Reduce the commitment with a short trial, limited audience, spending cap or staged contract. A safe-to-fail experiment makes failure tolerable through deliberate design.
- Match the process to the commitment. For a choice you can reverse at an acceptable cost, name an owner and a short decision deadline. For a hard-to-reverse choice, seek independent scrutiny as you compare alternatives and run a pre-mortem. Both categories require legal and safety checks.
- Specify the exit before entering. Record what would make you stop, who can authorize the reversal and how it will happen. Schedule a review. Without an exit mechanism, a reversible decision can become a permanent commitment through neglect.
Judge further research by the benefit it can deliver relative to the delay. Ask whether the next piece of information could change your choice enough to justify waiting. That’s the value of information, applied to the decision process itself.
04A worked example
Imagine an online retailer considering a shorter checkout page. The usual approval process takes six weeks. The team instead proposes a one-week trial: randomly assign 5% of eligible shoppers to the new layout, keep the existing payment system unchanged and retain the old page behind a switch.
What it looks like The team is bypassing a careful process to ship an unfinished idea. Some colleagues want the full review before any customer sees it.
What’s actually going on Two commitments have been bundled together. Permanently replacing checkout would affect every shopper. A limited trial with a tested rollback is a smaller decision. It can reveal usability problems while leaving adoption of the design optional. Random assignment also makes this an A/B test, though a short trial may produce too little data to establish a sales benefit.
What made it work The team first checks how payments behave and whether the page is accessible and secure. It tests the rollback, assigns someone to monitor failures and agrees to stop if payment errors rise. The owner has authority to restore the old page immediately. The trial’s scope, safeguards and exit are explicit.
Switching the page back cannot undo every customer’s frustrating experience. That residual cost is why the audience is limited. The door is acceptably reversible even though some consequences remain.
05When to reach for it
06When it misleads
- Reversible gets mistaken for harmless. You can delete a post, withdraw a policy or restore old software without undoing what people experienced. Reversing the action and reversing its consequences are different things.
- The rollback exists only on paper. An exit is unreliable when it depends on a backup nobody has restored or a cancellation right nobody has priced. Test the mechanism before relying on it.
- Someone else pays for the experiment. A company may recover cheaply while customers lose privacy or bear costs in time or money. Include the decision-maker’s balance sheet and the costs to everyone else affected when judging reversibility.
- Small commitments accumulate. Repeated temporary exceptions can create dependencies and shape what people habitually do and expect. Path dependence makes later reversal harder even when each initial step looked modest.
- The label becomes permission to skip thought. Reversible decisions still need enough care to bound the downside. Useful speed produces learning and keeps avoidable harm from being shifted onto others.
- A one-way door becomes an excuse to wait forever. Delay has costs too, and some opportunities expire. Use the extra deliberation to improve the analysis within a defined time limit. Decide what evidence you need and when you will make the call.
Also distinguish reversibility from flexibility before committing: Real options considers the value of retaining choices, such as waiting or expanding later. The door test is a simpler prompt for choosing how carefully to make the commitment.
07Roots
In the letter reporting Amazon’s 2015 results, Jeff Bezos addressed a problem that success itself could create. Amazon ran both an online store and a business selling computing services. As the organization grew, he worried that decisions would acquire heavier processes whether they needed them or not.
He gave the problem a physical image: a door you can walk back through versus one that closes behind you. He labeled consequential, hard-to-reverse choices Type 1 and reversible choices Type 2. The distinction was meant to stop a large company from treating routine experiments like permanent commitments. His concern was that excessive process would produce slowness, risk aversion and less invention.
In the following year’s letter, Bezos returned to decision speed and the importance of recognizing and correcting mistakes quickly. The memorable door image traveled into management and product development as a way to discuss both delegation and deliberation.
The importance of reversibility predates Bezos’s metaphor. Economists had already studied how irreversible investment and uncertainty make waiting valuable. His contribution was a portable management metaphor: first identify the kind of commitment, then choose the process.
08How solid is this?
This practical management heuristic is unvalidated as a decision rule. Research on irreversible investment supports the importance of preserving flexibility. Those findings leave open whether the two-door classification itself improves decision speed or quality.
09Connections
- Helps counter Information Bias
- Part ofReal Options, Option Value, Path Dependence
- Includes Safe-to-Fail Experiment
- See also Chesterton’s Fence, Sunk Cost Fallacy, Value of Information, A/B Testing, Irreversibility, Pre-Mortem, Via Negativa
+ 2 more in the list
10Origin and sources
Jeff Bezos popularized the distinction in Amazon’s 2015 shareholder letter, published in 2016, using Type 1 for hard-to-reverse decisions and Type 2 for reversible ones.
- [1]Bezos, J. (2016). 2015 Letter to Shareholders. Amazon.
- [2]Bezos, J. (2017). 2016 Letter to Shareholders. Amazon.
- [3]Dixit, A. K., & Pindyck, R. S. (1994). Investment under Uncertainty. Princeton University Press.
Suggest an edit· Updated 2026-10-02