Key Takeaways
Treat customer trust as the engine of long-term shareholder value
Start with the customer. Amazon’s central claim is that durable returns come from serving customers better over time, not from optimizing this quarter’s numbers at their expense. The practical test is whether a decision improves something customers value, such as price, selection, or convenience, while building a business that can sustain the improvement.
Trust has operating costs. Early on, a customer received a coffee-table book with a scratched cover. Amazon apologized, arranged a replacement, and let the customer choose whether to wait or keep the imperfect copy. That response cost more than simply shipping the damaged book, but it showed how the company wanted to earn trust. Kindle and Prime later applied the same logic at a much larger scale, accepting near-term investment to improve the experience.
The strongest version of this idea resembles a repeated-game strategy: a company earns future business by making the customer’s next decision easier and safer. That helps explain why a thoughtful remedy can matter beyond the cost of one order. The alignment, however, is a management belief, not an automatic law. Customer benefits can conflict with the interests of workers, suppliers, or communities, and apparent loyalty can conceal switching costs. A useful application therefore asks two questions: which customer outcome is improving, and who bears the cost of that improvement?
Turn principles into recurring mechanisms before the founder leaves the room
Culture needs machinery. A mechanism is a repeatable process that makes a desired behavior more likely, rather than a reminder to try harder. Amazon’s Leadership Principles were embedded in hiring, performance reviews, operating plans, and meetings so employees encountered them in decisions, not merely on posters.
Planning creates alignment. Teams draft OP1, their proposed annual operating plans, then adjust them in OP2 after the holiday results. Senior leaders select detailed priorities, called S-Team goals, and review progress throughout the year. The targets are deliberately aggressive: the authors say Amazon expected roughly three-quarters to be fully achieved. Senior compensation also leaned heavily toward company equity over several years, intended to reward durable value rather than local, short-term wins.
This is institutional memory in practical form. Organizations routinely lose their stated values when pressure rises, so recurring procedures can preserve priorities through leadership changes and rapid hiring. Yet a mechanism can turn into ritual if nobody checks whether it still produces the intended behavior. Aggressive targets may spur ambition, but they can also encourage concealment or metric gaming when missing a goal feels unsafe. The useful discipline is to review both the result and the mechanism: did the process improve decisions, or did people simply learn how to satisfy it?
Make candidates demonstrate their work, not just impress the interview room
Structure beats urgency. Amazon created the Bar Raiser process after rapid hiring weakened the quality of its software teams. In a fictional example, Green Corp hires Joe after interviewers trade positive impressions, a senior colleague forgets his reservations, and the group accepts vague praise as evidence. Joe leaves six months later, and the team must restart the search.
Collect evidence independently. Amazon’s process puts a trained interviewer, the Bar Raiser, in every interview loop. This person is neither the hiring manager nor recruiter and can veto a hire. Interviewers ask behavior-based questions, probe the candidate’s personal contribution, and record detailed feedback before seeing others’ views. A common probing structure is STAR: Situation, Task, Action, Result. The debrief then compares evidence against the job’s requirements and leadership expectations.
The method addresses familiar cognitive traps: urgency bias, groupthink, and the halo effect, in which one strong impression colors later judgments. Written evidence also makes disagreement easier to examine. The process is not automatically fair, though. Candidates differ in access to prestigious projects, comfort with self-promotion, and familiarity with interview conventions. A strong adaptation would define job-relevant criteria in advance, train interviewers to probe consistently, and periodically audit who advances at each stage. Structure can constrain bias, but it cannot decide whether the chosen criteria themselves are biased.
Give important work one focused owner and a team that can act independently
Coordination has a price. As Amazon grew, its teams became entangled through shared code, databases, approval chains, and competing priorities. Each dependency required people to coordinate before building. The company concluded that better meetings would not solve every delay; some dependencies had to be removed. In software, teams began owning separate services and interacting through defined interfaces rather than making changes directly in shared systems.
Focus the leader too. The early two-pizza team idea evolved into single-threaded leadership: one leader devotes their attention to a major initiative and has a team with the authority and skills to deliver it. Fulfillment by Amazon stalled while treated as a side project. Once Tom Taylor received dedicated focus and the ability to staff a team, the service launched in 2006 and grew into a major business.
This design joins organizational structure to technical architecture. Modular software boundaries reduce the need for engineers to negotiate every change, while clear ownership helps people know who is accountable for outcomes. The same principle can apply beyond software, but independence is not free: teams can duplicate work, neglect shared systems, or optimize against one another. Autonomy therefore needs explicit boundaries, shared priorities, and ways to resolve cross-team costs. The aim is not to eliminate communication, but to reserve it for coordination that genuinely creates value.
Make leaders read the argument before they hear the presenter
Write to clarify. Amazon replaced PowerPoint at senior-team meetings with a six-pager, a concise narrative that presents a proposal or business review in connected prose. Meetings begin with about 20 minutes of silent reading, then move to questions and discussion. The format makes the team explain its reasoning, evidence, and trade-offs before relying on a polished speaker to make the case.
Density changes the discussion. The authors compared typical narrative pages, at 3,000 to 4,000 characters, with Amazon presentations averaging 440 characters per slide. They estimated the narrative carried seven to nine times more information per page. Charts and appendices remain useful, but the document must make sense on its own. The audience can annotate, challenge assumptions, and discuss the same complete argument instead of reconstructing it from slides and spoken remarks.
The format reduces dependence on charisma and gives complex arguments room to show how evidence connects. It also shifts effort upstream: the writer must synthesize before the room spends expensive executive time. This echoes a basic finding from cognitive science: people understand relationships more easily when relevant information can be revisited, rather than held in working memory during a presentation. Still, prose is not neutral. It can privilege fluent writers, and a dense memo can bury a weak claim as effectively as a glossy slide. Clear standards, editing support, and accessible data visualizations matter.
Write the customer promise before committing engineers and money
Begin at the finished experience. Working Backwards is Amazon’s product-development approach of defining the customer outcome first, then reasoning back toward what must be built. Its main tool is the PR/FAQ: a short press release describing the product as if it were ready, followed by questions that test customer experience, cost, technical feasibility, and dependencies.
Kindle took shape this way. Early proposals focused on market size and financial projections, but Jeff Bezos kept asking how the service would work for customers. The resulting device combined an e-bookstore with wireless buying and downloading, an E Ink screen, and a larger catalog than the competing Sony reader at launch: 90,000 titles compared with about 20,000. The point was not to predict every detail perfectly. It was to expose hard questions while changing direction was still relatively cheap.
This is a form of design reasoning that makes the desired outcome concrete enough to interrogate. A customer-facing promise can reveal contradictions between features, price, and delivery that a market-size spreadsheet may obscure. The risk is treating an imagined press release as proof of demand. A persuasive hypothetical customer quote is still a hypothesis, not customer evidence. The method is most powerful when paired with research, prototypes, and explicit uncertainty, so a team can distinguish a crisp narrative from a validated need.
Manage the actions that cause results, not results alone
Outputs report what happened. Revenue and profit matter, but teams cannot directly set them in a reliable way. Amazon instead emphasizes controllable input metrics: activities that can be changed and are expected to influence customer outcomes. Its retail teams first counted new product pages, assuming more pages meant better selection. Sales did not rise as expected, while unwanted inventory accumulated.
Refine the measure. The selection metric evolved from pages created to pages viewed, then to customer views of items in stock and ready for quick shipping. This more useful measure helped teams identify products customers actually wanted. The Weekly Business Review, or WBR, examines trends, anomalies, and customer stories to find causes and assign action. One customer complaint about repeated card authorizations leading to a declined medicine purchase helped expose a software defect and prompted a way for customer service to flag problematic products.
The approach resembles control theory: a system improves when decision-makers can observe meaningful inputs, understand variation, and adjust causes rather than react to outcomes alone. But measures change behavior, so every input metric needs scrutiny. Page views, for example, can rise while customer value falls. Combining quantitative trends with a concrete customer account can expose what averages hide, though anecdotes can also be unrepresentative. The practical safeguard is to ask whether the metric reflects a real customer experience, whether it can be independently checked, and what behavior it rewards.
Let customer evidence justify patient bets, not sunk-cost loyalty
Prime made the long bet. In 2004, Amazon saw that free shipping attracted customers but its existing network could not deliver fast shipping cheaply. The company chose to invest in a better system, then launched a paid membership with free two-day shipping in February 2005, less than four months after the urgent launch directive. By 2018, Prime had passed 100 million paid members globally.
Persistence is not proof. Amazon Unbox, its early movie-download service, was slow, difficult to install, limited to some Windows computers, and burdened by playback problems. The Fire Phone also failed to solve a compelling customer problem, despite extensive engineering. It was discontinued in 2015. The authors’ distinction is crucial: a disciplined process can improve decisions, but cannot guarantee success. Keep testing whether the original customer need remains important, and change course when the evidence weakens.
These cases illustrate a portfolio logic: a company can tolerate failures when the cost of experiments is bounded and a few successful bets can repay many misses. In behavioral economics, sunk-cost bias describes the tendency to continue mainly because resources have already been spent. Amazon’s language of patience can resist short-term panic, but it can also be used to excuse a failing project. A useful test is whether new evidence strengthens the customer case and changes the plan, not whether leaders can tell a more convincing story about past investment.
Give customers basic tools, and they may reveal a business you missed
Follow unexpected use. Amazon’s early affiliate program supplied product information to outside websites. In 2002, it let technically skilled affiliates receive the information in a format they could customize themselves. Developers built unexpected tools and interfaces, and their requests showed that the opportunity was larger than a way to sell Amazon products.
A new customer emerged. Amazon opened its web services to software developers, who could use tools and data through defined interfaces. More than 25,000 developers had enrolled by 2003. This early work did not yet resemble modern cloud computing, but it helped reveal demand for reusable computing services. Amazon had experience operating large-scale systems, and software developers wanted infrastructure they could use without building it all themselves. AWS grew from these small experiments into a major business, reporting $35 billion in revenue in 2019.
The story captures a platform dynamic: when a company offers reliable building blocks, customers may combine them in ways the provider did not anticipate. Their behavior becomes a source of product discovery, not just a response to a finished product. The strategy also requires restraint. Opening valuable capabilities can create security, reliability, and governance obligations, and a lively developer community does not automatically imply a sustainable business. The useful signal is not novelty alone, but repeated use that solves a meaningful problem for people beyond the original customer group.
Analysis
The book’s most useful contribution is not a single Amazon technique. It is the way the techniques reinforce one another. Customer focus defines the aim, written proposals force teams to make the aim explicit, focused ownership gives someone the authority to act, and input metrics make it possible to see whether the work is changing the customer experience. Hiring and planning processes then help reproduce those habits as the company grows. The result is a management system that tries to replace heroic dependence on a founder with repeatable organizational behavior.
Its case histories also show that innovation is less tidy than a framework diagram suggests. Prime took shape through years of shipping experiments and a sudden compressed launch. AWS grew from a tool for affiliates, while Unbox and Fire Phone demonstrate that rigorous effort can still misread demand. These stories strengthen the argument when they show revision and uncertainty, but they also invite survivor-bias questions: successful companies can explain past choices as principled discoveries after the outcome is known. Process does not prove that a bet was wise at the time.
The book is deliberately a guide to Amazon’s internal practices, not a complete assessment of Amazon as an institution. That boundary matters. A customer-first operating system can improve delivery speed and product quality without settling questions about labor conditions, market power, seller dependence, or broader social costs. The authors acknowledge that these concerns are outside their scope. Readers should therefore borrow the methods without treating commercial success as evidence that every stakeholder benefits.
The deepest practical lesson is conditional: mechanisms are powerful because they shape attention and incentives, but they are not substitutes for judgment. A six-page memo can expose fuzzy thinking or make it harder to notice. A metric can illuminate a process or distort it. Autonomy can unleash builders or fragment an organization. Amazon’s methods work best as revisable instruments, tied to a clearly defined customer need and subject to honest review of both intended and unintended effects.
Review Summary
Working Backwards offers an insider's view of Amazon's culture and practices. Readers appreciate the practical insights into Amazon's unique approaches, such as the "bar raiser" hiring process, six-page narratives, and customer-obsessed product development. The book is praised for its detailed examples of Amazon's successes and failures, including the development of Kindle, Prime, and AWS. While some readers find the content eye-opening and applicable to their own businesses, others criticize it for being too laudatory of Amazon and glossing over worker treatment issues.
People Also Read
Glossary
Mechanisms
Repeatable behavior-shaping processesAt Amazon, mechanisms are recurring processes that make desired behaviors more likely and help prevent known problems from recurring. Examples include annual operating plans, goal reviews, and hiring procedures. They differ from exhortations or good intentions because they alter how work is organized, measured, or decided.
Bar Raiser
Independent hiring quality controlThe Bar Raiser is both Amazon’s structured hiring process and the trained interviewer who participates in each interview loop. This person is separate from the hiring manager and recruiter, helps keep evaluation evidence-based, coaches interviewers, and has authority to block a hire that does not meet the company’s standard.
single-threaded leadership
One leader, one major focusSingle-threaded leadership assigns a leader to focus on one major initiative, supported by a team with the authority, skills, and resources to deliver it. The team has clearly defined ownership and aims to minimize dependencies on other teams. The model grew out of Amazon’s experiments with smaller autonomous teams.
six-pager
Concise narrative for decisionsA six-pager is a short written narrative used to propose, explain, or review a business idea. Amazon meetings using one begin with silent reading, followed by discussion. The format asks teams to connect evidence and reasoning in prose, rather than relying on a slide presentation and a speaker to supply missing context.
Working Backwards
Design from customer outcomeWorking Backwards is Amazon’s approach to developing products by first defining the intended customer experience and then reasoning toward the product, technology, costs, and organization needed to deliver it. Teams use written documents to expose assumptions and constraints before committing substantial development resources.
PR/FAQ
Product promise and hard questionsThe PR/FAQ is a product-development document used in Working Backwards. Its brief press release describes the proposed product and its customer benefit as if it were ready to launch. The accompanying frequently asked questions examine details such as usability, economics, technical feasibility, risks, and dependencies.
controllable input metrics
Manageable drivers of resultsControllable input metrics measure activities or conditions a team can directly influence and expects to affect business outcomes. Amazon examples include product availability and delivery speed. They complement output metrics, such as revenue, which indicate results but are harder for teams to control directly.
Weekly Business Review (WBR)
Weekly review of business signalsThe WBR is a recurring meeting in which teams review business metrics, trends, exceptions, and relevant customer accounts. Owners are expected to understand their measures, explain meaningful variances, and identify actions. Amazon uses the review to investigate causes and improve operations, rather than simply report results.
Amazon flywheel
Self-reinforcing growth cycleThe Amazon flywheel describes a reinforcing cycle in its retail business. Better customer experience attracts more traffic, which can attract more sellers and broaden selection. Greater selection can improve the experience, while growth can lower costs and support lower prices. Improvements in one part of the cycle can strengthen the others.
Correction of Errors (COE)
Root-cause learning processCorrection of Errors is Amazon’s written process for examining a significant problem, explaining what happened, and tracing contributing causes, often by asking why repeatedly. The goal is to find and address the underlying process weakness so the same failure is less likely to recur, rather than stopping at blame or a quick fix.
FAQ
What's Working Backwards about?
- Insider's Perspective: The book offers a deep dive into Amazon's corporate culture, focusing on its unique operational strategies and leadership principles.
- Customer Obsession: It highlights Amazon's core philosophy of prioritizing customer needs, aligning long-term business interests with customer satisfaction.
- Practical Frameworks: The authors introduce methodologies like the Working Backwards process, which starts with the desired customer experience to drive innovation.
Why should I read Working Backwards?
- Learn from Experience: Authors Colin Bryar and Bill Carr share insights from their extensive careers at Amazon, offering lessons applicable to any organization.
- Actionable Strategies: The book provides concrete strategies and principles that can be implemented in various business contexts, making it a practical guide for leaders.
- Understanding Innovation: It reveals how Amazon maintains its innovative edge, inspiring similar practices in other companies.
What are the key takeaways of Working Backwards?
- Four Core Principles: Amazon's culture is built on customer obsession, long-term thinking, eagerness to invent, and operational excellence.
- Leadership Principles: The book outlines Amazon's 14 Leadership Principles, which guide decision-making and behavior within the company.
- Working Backwards Process: This process involves writing a press release and FAQ before product development, ensuring a customer-centric approach.
What is the Working Backwards process in Working Backwards?
- Customer-Centric Approach: It starts with defining the customer experience and works backwards to clarify what needs to be built.
- PR/FAQ Tool: Utilizes a press release and FAQ document to outline the product's value proposition and address potential challenges.
- Iterative Refinement: Encourages multiple drafts and revisions, fostering critical thinking and thorough analysis before product development.
How does Amazon's hiring process work according to Working Backwards?
- Bar Raiser Program: A unique process ensuring high hiring standards, with a designated Bar Raiser having veto power in interviews.
- Focus on Culture Fit: Emphasizes finding candidates who align with Amazon's Leadership Principles, contributing positively to the company culture.
- Structured Interviews: Uses behavioral questions to assess candidates' past experiences and alignment with Amazon's values.
What are Amazon's Leadership Principles as described in Working Backwards?
- Guiding Framework: The 14 Leadership Principles influence decision-making and behavior, from hiring to performance evaluations.
- Examples of Principles: Includes "Customer Obsession," "Invent and Simplify," and "Insist on the Highest Standards."
- Continuous Evolution: Regularly reviewed and refined to adapt to new challenges, ensuring relevance as the company grows.
How does Amazon manage its metrics according to Working Backwards?
- Input vs. Output Metrics: Focuses on controllable input metrics, emphasizing activities that drive business results.
- Weekly Business Review (WBR): Regularly reviews key metrics, ensuring accountability and alignment with company goals.
- Data-Driven Decisions: Metrics inform decision-making, with emphasis on understanding factors driving performance.
What is the significance of narratives at Amazon as mentioned in Working Backwards?
- Replacing PowerPoint: Written narratives replace presentations to enhance clarity and depth of thought in meetings.
- Eerie Silence: Meetings begin with silence as attendees read the narrative, leading to informed discussions.
- Narrative Information Multiplier: Allows for greater information density, enabling deeper insights and effective communication.
What are some challenges Amazon faced during its growth according to Working Backwards?
- Coordination Issues: Growth led to increased complexity in team coordination, causing delays and inefficiencies.
- Dependency Management: Struggled with technical and organizational dependencies, prompting new structures and processes.
- Transition to Autonomy: Shifted to separable, single-threaded teams to reduce dependencies and enhance innovation speed.
What is the Amazon flywheel as described in Working Backwards?
- Virtuous Cycle: Improvements in customer experience drive traffic, attracting more sellers and enhancing customer experience further.
- Input Metrics: Emphasizes selection and price, which influence output metrics like revenue and growth.
- Continuous Improvement: Ensures every business aspect is interconnected, contributing to overall success.
What are some examples of successful inventions at Amazon mentioned in Working Backwards?
- Kindle: Revolutionized reading and established Amazon as a leader in digital media.
- Amazon Prime: Transformed customer expectations around shipping and convenience, driving company growth.
- AWS: Established Amazon as a leader in cloud computing, providing scalable solutions for businesses.
What are the best quotes from Working Backwards and what do they mean?
- “We make money when we help customers make purchase decisions.”: Highlights Amazon's philosophy of prioritizing customer needs for long-term profitability.
- “Speed matters in business.”: Emphasizes timely decision-making and calculated risks to stay competitive.
- “We don’t let defects travel downstream.”: Reflects commitment to quality and proactive issue resolution before affecting customers.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.