Business thinker

Jeff Bezos's working backwards, one-way doors and Day 1

Updated 4 November 2026

Not affiliated with or endorsed by Jeff Bezos or their companies.

Jeff Bezos, Amazon's founder, described his working principles in letters to shareholders. This page covers three: working backwards from what the customer needs, which Amazon turns into a PR/FAQ written before anything is built; sorting decisions into one-way and two-way doors; and staying a Day 1 company. This page explains each from the 2008, 2015 and 2016 letters, with the PR/FAQ format as Amazon itself describes it.

Working backwards from the customer

In the 2008 letter, Bezos contrasts working backwards from customer needs with a "skills-forward" approach, where a company asks what else it can do with what it is already good at. He calls skills-forward useful, but says a company that relies on it alone is never pushed to build new skills. Working backwards often demands new skills. His example is Kindle: Amazon had never built a hardware device, and hired hardware engineers rather than change the vision to fit the skills it had.

The 2016 letter adds a warning about proxies. Market research and customer surveys can stand in for real understanding of the customer, which is especially risky when inventing products. Bezos writes that he isn't against surveys, but the owner of a product has to understand the customer and have a vision; research then helps find blind spots.

The PR/FAQ: write the press release first

Amazon's 2024 shareholder letter, written by Andy Jassy, says that before it starts coding a service or feature, Amazon writes a press release and an FAQ. The press release is meant to make sure the proposal is remarkable to customers. The FAQ forces hard questions: which customers will use it, what they'll like most and what will disappoint them, why the launch line is drawn where it is, why it beats current alternatives, how to think about pricing, and why the architecture is what it is.

An excerpt from Working Backwards by former Amazon employees Colin Bryar and Bill Carr, published on Amazon's website, gives the format. The press release is a few paragraphs, always under one page; the FAQ is five pages or less. Reviewers often ask "so what?": if the release doesn't describe something meaningfully better (faster, easier, cheaper) than what exists, or a step change in the customer experience, it isn't worth building. In a review, everyone reads the document first, and the most senior people tend to speak last. The authors write that most PR/FAQs never become products, and treat that as a benefit: it keeps build resources for the best ideas.

Take a fictional bookkeeping firm for restaurants that has a script for reading supplier invoices. A skills-forward pitch would be to sell the script as it is. A PR/FAQ starts from the owner who only learns food costs at month end, writes the release about seeing each dish's cost every Monday, and then has to answer in the FAQ how many owners have recipes written down per dish.

The PR/FAQ is an internal thinking document. A news release for journalists is a different task, and so is a memo that argues for one decision; the library's press release and decision memo skills cover those.

One-way and two-way doors

The 2015 letter names one-size-fits-all decision making as a common pitfall for large organizations. Some decisions are consequential and irreversible or nearly irreversible: one-way doors. Bezos writes that these must be made methodically, carefully and slowly, with deliberation and consultation, and calls them Type 1 decisions. Most decisions are reversible: two-way doors. If a Type 2 decision turns out poorly, you can go back through the door, so these should be made quickly by individuals with high judgement or small groups.

The risk he points to is that, as organizations grow, they use the heavy Type 1 process for most decisions, including many Type 2 ones. The result, the letter says, is slowness, unthoughtful risk aversion, too little experimentation and less invention. A footnote covers the opposite habit: companies that routinely use the light process for Type 1 decisions die out before they grow large.

For a fictional language school with four decisions waiting on the same weekly meeting, a new homepage headline and a test onboarding email are two-way doors that one person can decide this week; a three-year studio lease and a move to annual-only plans are one-way doors that need a written case and consultation.

Day 1 and high-velocity decisions

The 2016 letter opens with a question from an all-hands meeting about what Day 2 looks like. Bezos describes Day 2 as stasis, then irrelevance and decline, which is why it is always Day 1. His starter pack for staying in Day 1 is customer obsession, a skeptical view of proxies (including process becoming the thing instead of the result), the eager adoption of external trends, and high-velocity decision making.

On decisions, the letter adds three points to the two-way door idea. Most decisions should probably be made with somewhere around 70% of the information you wish you had; if you wait for 90%, you're probably being slow in most cases, and either way you need to be good at quickly recognizing and correcting bad decisions. When there is conviction but no consensus, use the phrase "disagree and commit", which Bezos says applies to the boss as well. And when teams have truly different objectives, escalate early; otherwise the decision goes to whoever has more stamina.

How to apply it to your own decisions

1. Before building something new, write who the customer is and what need they have, in one sentence. If the idea starts from something you already have, find the customer need first.
2. Write a press release of under one page from the customer's side, then ask "so what?" against what customers use today.
3. Write the FAQ with the hard questions: who uses it first, what will disappoint them, why the launch line is here, pricing, cost. Mark what you don't know.
4. List the decisions waiting on you. For each, write what it would cost to undo.
5. Give each reversible decision one decider and a short deadline; give each hard-to-reverse one a written case and the people to consult.
6. Where people disagree on a shared goal, ask for disagree and commit; where the goals themselves differ, escalate now.

What the sources say, and what they don't

Working backwards, one-way and two-way doors, the 70% line, disagree and commit and Day 1 come from Bezos's own shareholder letters (2008, 2015, 2016). The PR/FAQ page limits, the so-what question and the review practice come from Bryar and Carr's excerpt on Amazon's website, and the before-coding practice and FAQ questions from Andy Jassy's 2024 letter; they describe Amazon's process, not words of Bezos. The restaurant and language school examples are ours.