One-Way and Two-Way Doors (Amazon’s 2015 Letter)

The situation

By the time Bezos wrote his letter on 2015, Amazon was very large. The letter reports that Prime’s two-day delivery selection had grown from 1 million to more than 30 million items and that membership grew 51 percent that year; that close to 50 percent of units sold on Amazon came from third-party sellers; and that AWS was used by more than a million customers and had announced 722 significant new features and services in 2015, a 40 percent increase over the year before.

The letter’s explicit concern was whether a company of that size could keep the qualities associated with start-ups. Bezos framed the goal as combining the advantages of scale with the “nimbleness, and risk-acceptance mentality normally associated with entrepreneurial start-ups,” and acknowledged that there were “some subtle traps that even high-performing large organizations can fall into.”

What the letter argues

He named the first trap “one-size-fits-all decision making.” Then came the passage that made the letter famous.

Some decisions, he wrote, “are consequential and irreversible or nearly irreversible – one-way doors,” and they must be made slowly, with great deliberation and consultation, because if you walk through and dislike what you find, you cannot get back. He called these Type 1 decisions.

“But most decisions aren’t like that – they are changeable, reversible – they’re two-way doors.” If a Type 2 decision turns out badly, you can reopen the door and go back through. Hence: “Type 2 decisions can and should be made quickly by high judgment individuals or small groups.”

The problem, he argued, is drift. “As organizations get larger, there seems to be a tendency to use the heavy-weight Type 1 decision-making process on most decisions, including many Type 2 decisions. The end result of this is slowness, unthoughtful risk aversion, failure to experiment sufficiently, and consequently diminished invention.”

A footnote adds the opposite failure, with characteristic dryness: “Any companies that habitually use the light-weight Type 2 decision-making process to make Type 1 decisions go extinct before they get large.”

The decision it carried

The letter announced no product. Its decision was about management: Amazon would treat reversibility as the main factor in deciding how a decision should be made, and would actively push reversible decisions down to individuals and small teams. Publishing the idea to shareholders made it a public principle against which the company’s own bureaucracy could be measured.

The alternatives are the defaults of most large companies: a single approval process for all significant decisions, or decisions escalated by size of budget rather than by reversibility.

What happened

The distinction spread far beyond Amazon. “One-way door” and “two-way door” became common vocabulary in product and engineering teams across the technology industry, used in design reviews, launch decisions and planning. Bezos returned to the theme in his 2016 letter, arguing that large companies must keep decision-making velocity high and that most decisions should be made with somewhere around 70 percent of the information one would like to have.

Inside Amazon, the same letter illustrates the principle with the company’s own history: it notes that Amazon “took two big swings and missed – with Auctions and zShops” before Marketplace succeeded. Reversible bets, made quickly and abandoned when they failed, were part of how the eventual successes were found.

How to read it as a product manager

This is perhaps the most directly usable idea in any document on this page.

Classify before you deliberate. Before a meeting about a decision, ask whether it is a one-way or two-way door. If it is two-way, the right question is often who should decide quickly, not how many people should review it.

Use reversibility to unblock teams. Many product decisions (copy, layouts, feature flags, small experiments) are cheap to undo. Treating them with the caution appropriate to pricing changes or platform choices slows everything down.

Watch for hidden one-way doors. The framework is only as good as the classification. A feature can be technically reversible but create user expectations, data dependencies or public commitments that make removing it costly. Google’s experimental Inbox app, closed in 2019 after users had come to rely on it, is an example discussed on this site. Ask what would actually happen if you reversed the decision six months from now.

Remember the footnote. Speed is not a universal virtue. Choosing an architecture, entering a market, signing a long contract or making a public promise can be one-way doors that deserve the slow process.

What is contested

The framework is widely praised and rarely criticised in principle, but its application is debated. Critics note that organisations can label decisions “two-way doors” to avoid scrutiny, and that the people affected by a reversible decision may experience it as anything but reversible. The letter also presents Amazon’s culture in the best light, as any shareholder letter does; it does not discuss the costs of fast decisions for employees or partners.

The decision, examined closely

What triggered it

Amazon had become one of the largest companies in the world, with tens of thousands of new employees each year, and Bezos was openly worried about the slow, cautious habits that large organisations develop. The letter names this as its central question: whether Amazon could keep a start-up’s nimbleness at its scale.

The constraints

A company as large as Amazon cannot have its chief executive approve every decision, and it cannot let every team make irreversible commitments on its own. Any solution had to be simple enough to be used by thousands of managers without central supervision.

What the author was optimising for

The letter optimises for the rate of invention. Its logic is that invention requires many experiments, experiments require fast decisions, and fast decisions are safe only when they can be undone. Reversibility is the variable that connects speed and safety.

Second-order effects

The vocabulary escaped Amazon and became part of how product and engineering teams across the industry talk about decisions. It gave people at every level a polite way to challenge unnecessary process (“isn’t this a two-way door?”). It may also have encouraged a habit of shipping quickly and withdrawing later, which has costs for users who adopt a product that is then removed.

The counterfactual

Without an explicit framework, Amazon would still have had fast and slow decisions, but the boundary would have been set by habit and seniority rather than by the nature of the decision. The letter’s contribution was to make the boundary visible and discussable, which is often the most a management idea can do.

Sources and limitations

Quotations were taken from Amazon’s own PDF of the letter. The business figures are those stated in the letter. The 2016 letter is summarised rather than quoted extensively. The exact publication date of the 2015 letter is approximate, based on the date of Amazon’s published file. The assessment of the framework’s spread and limits is the author’s interpretation.