Working Backwards

TL;DR: Amazon’s operating system is a set of mechanisms — PR/FAQ, single-threaded leadership, two-pizza teams — designed to scale invention by eliminating the need for coordination. The book’s most quotable line is also its thesis: “Good intentions don’t work. Mechanisms do.”

What it means

Bryar and Carr were inside Amazon during the years when it was figuring out how to stay startup-fast while becoming planet-sized. The book’s thesis is simple and corrosive to most management writing: Amazon doesn’t rely on culture or values to drive behavior. It builds processes that make the desired behavior the path of least resistance. This is process-power by design, not by accident. If you want people to start with the customer, don’t put it in the values deck. Make them write a customer-facing press release before they’re allowed to write any code.

The PR/FAQ process is the most exportable idea in the book. Before building anything, you write the press release announcing the finished product, then a six-page FAQ answering tough questions from both customers and internal stakeholders. This forces clarity on who the customer is, what problem you’re solving, and why it matters — before a single line of code gets written. It’s working-backwards-method as institutional discipline rather than personal taste, and it kills more bad projects per dollar than any other process Amazon has ever shipped.

The organizational architecture is equally deliberate. Single-threaded leadership means one leader owns one project with full focus — no committee decisions, no shared priorities, no “and they also help with the other thing.” Two-pizza teams are small autonomous units evaluated by a fitness function (a measurable output metric), not by how well they collaborate with other teams. Both are designed to make coordination unnecessary rather than easier, which is the inverse of how every other large company solves the same problem.

The argument

Customer obsession over competitor obsession. Amazon’s most counterintuitive bet is ignoring competitors. Competitor-focused companies wait for the market to move and then react. Customer-focused companies invent ahead of demand. This is why Amazon built AWS, Kindle, and Prime before obvious market signals existed — because the customer’s pain was visible even though no competitor was solving it. Customer obsession is the upstream version of counter-positioning: you don’t have to outmaneuver the incumbent if you’re solving a problem the incumbent isn’t even tracking yet.

Eliminate communication, not encourage it. “If you want to be a place where builders can build, eliminate communication, not encourage it.” This sounds heretical, but the logic is sound: communication is overhead. Loosely coupled services connected via APIs let teams move independently and ship without checking in. This is the technical architecture that enables two-pizza teams to function. The “more meetings, more alignment” instinct is exactly the opposite of what works at scale.

Long-term equity compensation. Amazon historically paid below-market cash and compensated with equity that vested over years. This filters for people who believe in the long game and aligns incentives toward durable value creation rather than quarterly results. The mechanism here is the same as everything else Amazon does: don’t ask people to think long-term, pay them in a way that makes long-term the only rational frame.

Inventor vs. fast-follower. Amazon explicitly chose the inventor path over fast-following, especially for digital products. Fast-following works in mature markets with established demand. Invention creates the markets where power-progression begins — at the origination stage where counter-positioning and cornered-resource are still available. Fast-followers can never have those Powers; they were locked in before the followers showed up.

Mechanisms over intentions. The recurring theme: Amazon treats organizational design as engineering. Every desired behavior gets a mechanism — a process, a metric, a review cadence, a structural constraint. This is how you scale culture without diluting it. Most companies try to scale culture by hiring “culture fit” candidates and writing better values decks. Amazon scales culture by making it impossible to do the wrong thing inside their systems. The distinction matters more the bigger you get.