Amazon Leadership Principles Interview Questions: All 16 With STAR Answers (and the 3 That Reject Most Candidates)

Every Amazon interviewer walks into your loop with two or three of the 16 Leadership Principles assigned to them, pulls questions from an internal bank organized by principle, and scores your answers in a written debrief before anyone discusses the hire. Most candidates prepare for Amazon leadership principles interview questions by memorizing the list. That fails, because the list is public and your interviewer knows you have read it. What gets people hired is a bank of 8 to 10 real stories, quantified, mapped to the principles, and rehearsed against the follow-up questions Amazon trains its interviewers to ask.

This guide covers all 16 principles with 48 real questions and a senior-level STAR answer for each principle. It also names the three principles where strong candidates fail most often, because knowing where the bar is highest changes how you spend your prep time.

How Amazon actually scores you

Four things about the process change how you should prepare.

Each interviewer owns specific principles. A loop of 4 to 6 interviewers divides the principles among them. Nobody asks about all 16. Each interviewer asks 2 to 3 behavioral questions from Amazon's internal question bank, then spends most of the time on follow-ups.

The follow-ups are the interview. First answers are assumed to be rehearsed. Expect three levels of probing: what did you do, why that way, what would you do differently. If your story is thin, level three exposes it.

One interviewer is a Bar Raiser. They come from outside the hiring team, they are trained to keep the bar above the team's current average, and they chair the debrief where the decision is made. They probe hardest on principles.

That round has its own mechanics, and the veto everyone describes is not what Amazon actually documents. Full breakdown in our Amazon Bar Raiser interview guide.

"We" answers score poorly. Amazon scores what you did. Every time you say "we shipped it," the interviewer writes down that they still do not know what you did. Use "I" and name your specific decisions.

STAR, the Amazon version

You know STAR: Situation, Task, Action, Result. The Amazon version has different proportions. Situation and Task get two sentences combined. Action gets 60 percent of your time, told in "I" statements. Result gets numbers: revenue, cost, time saved, error rates, adoption. A result without a number is an opinion.

One more Amazon habit: end with what you learned or changed afterward. Learn and Be Curious is a scored principle, and a one-line "since then, I..." closer feeds it for free.

The 3 principles that reject the most senior candidates

Across the mock interviews we run at Skillora, three principles produce the weakest answers from otherwise strong candidates:

• Customer Obsession, because senior engineers and finance operators reach for internal-stakeholder stories. Amazon wants the end customer, and "my stakeholder was the product team" reads as a dodge.
• Ownership, because candidates describe doing their job well. Ownership stories must cross a boundary: you fixed something that was not yours to fix.
• Have Backbone; Disagree and Commit, because both halves must be present. Candidates tell a disagreement story and skip the commit, or tell a commit story with no real disagreement. The principle is the sequence.

If your prep time is limited, build your best stories for these three first.

All 16 principles, with questions and answers

1. Customer Obsession

Leaders start with the customer and work backwards. The interviewer wants evidence that you traded something real, time, revenue, or your own roadmap, for a customer outcome.

Questions:

• Tell me about a time you went beyond what a customer asked for.
• Describe a decision where customer impact and business metrics pointed in opposite directions.
• Tell me about a time a customer asked for one thing but needed another.

A STAR answer that works: Situation: Our largest enterprise account asked for a custom export format three weeks before their audit. Task: The roadmap had no room, and the request affected one customer. Action: I interviewed their compliance lead, found the audit requirement behind the request, and shipped a generic audit-export instead of their custom spec. I cut two planned features to do it. Result: The account renewed at $340K, and 14 other customers adopted the audit export within a quarter.

The trap: telling a story where "the customer" is your product manager. Internal stakeholders do not count.

2. Ownership

Leaders act on behalf of the entire company, beyond their own team. The interviewer is listening for the moment you crossed a boundary you did not have to cross.

Questions:

• Tell me about a time you took on something outside your job description.
• Describe a problem you fixed that your team did not own.
• Tell me about a long-term bet you made that cost you short-term.

A STAR answer that works: Situation: Deploy failures in a sister team's service were paging my on-call rotation twice a week. Task: Their backlog had the fix scheduled two quarters out. Action: I wrote the runbook fix myself, got their tech lead to review it, and then proposed a shared deploy-gate standard to both directors. Result: Cross-team pages dropped 80 percent, and the deploy gate became the org standard for 9 services.

The trap: describing excellence inside your own lane. Doing your job well is the baseline, not ownership.

3. Invent and Simplify

Leaders find new solutions and remove complexity. Simplification stories often score better than invention stories, because they are harder to fake.

Questions:

• Tell me about a process you simplified. What did you delete?
• Describe the most creative solution you have shipped under a constraint.
• Tell me about a time your invention failed.

A STAR answer that works: Situation: Our quarterly planning consumed three weeks across 6 teams and produced a deck nobody reopened. Task: As the senior engineer in the room, I had no authority to change the process, only to propose. Action: I replaced the deck with a one-page template per team, six boxes, hard word limits, and ran the pilot myself with two teams before asking for the org. Result: Planning dropped to four days. The template spread to 14 teams, and two VPs adopted it for their own reviews.

The trap: confusing "I built something complex" with invention. Amazon scores the removal of complexity, not the addition of it.

4. Are Right, A Lot

Leaders have strong judgment and seek out disconfirming views. The interviewer wants your decision-making process, not your batting average.

Questions:

• Tell me about a high-stakes decision you made with incomplete data.
• Describe a time you changed your mind after new evidence.
• Tell me about your biggest wrong call. How did you catch it?

A STAR answer that works: Situation: We had to pick between rebuilding our billing system or extending it, with a hard compliance deadline in five months. Task: I owned the recommendation to the VP. Action: I wrote both options as one-pagers, gave the rebuild case to its strongest opponent to argue, and set a kill criterion in advance: if the prototype missed two of its first three milestones, we extend instead. It missed two. Result: We extended, hit the compliance date with six weeks of margin, and rebuilt the following year with a real runway.

The trap: claiming you were right because the outcome was good. Amazon scores the process. A good outcome from a coin flip is still a coin flip.

5. Learn and Be Curious

Leaders keep improving and act on new knowledge. The interviewer wants recency: what you learned lately and what changed because of it.

Questions:

• What have you learned in the past six months that changed how you work?
• Tell me about a time you were the least knowledgeable person in the room.
• How do you decide what to learn next?

A STAR answer that works: Situation: Our team started getting LLM-related feature requests and I had no hands-on depth. Task: I needed real judgment, not vocabulary, within a quarter. Action: I built an internal prototype on evenings for three weeks, wrote up the failure modes I hit, and ran a working session for the team on what the prototypes could not do. Result: We rejected two feature requests that would have failed, shipped one that worked, and my write-up became the team's evaluation checklist.

The trap: naming a certification and stopping. Learning that did not change a decision is trivia.

6. Hire and Develop the Best

Leaders raise the performance bar with every hire and take coaching seriously. Senior candidates get scored on people they grew, not tasks they finished.

Questions:

• Tell me about the best person you ever hired. How did you find and close them?
• Describe someone you coached from underperforming to promoted.
• Tell me about a hiring mistake and what you changed after it.

A STAR answer that works: Situation: A strong mid-level engineer on my team was stuck: two promo cycles, no movement. Task: As their lead, I owned the case. Action: I diagnosed the gap as visibility, not skill, gave them our highest-ambiguity project, reviewed their design docs before wide circulation, and put them in front of the director monthly. Result: Promoted in the next cycle, and they now run the subsystem. My promo-prep checklist from that case is used across the org.

The trap: taking full credit for someone else's growth. Score comes from the mechanism you built, not from claiming their success.

7. Insist on the Highest Standards

Leaders enforce standards that look unreasonably high from outside. The interviewer wants a story where quality cost you something and you paid it.

Questions:

• Tell me about a time you refused to ship.
• Describe a standard you set that your team initially resisted.
• Tell me about a time you accepted lower quality. Was it right?

A STAR answer that works: Situation: Two days before a marquee launch, error rates in one flow were 3x our target, but only for 2 percent of users. Task: Launch owner wanted to ship; I owned the quality call. Action: I held the launch, published the exact bar we were missing with the data, and split the release so the healthy 98 percent flow shipped on time while we fixed the broken segment. Result: Full launch went out nine days later with errors under target. The split-release pattern became our default for risky launches.

The trap: perfectionism stories. Blocking a launch over something users never noticed scores negative. The standard must map to real impact.

8. Think Big

Leaders propose directions that sound slightly unreasonable at first. The interviewer wants scope: did your idea outgrow your team?

Questions:

• Tell me about your most ambitious proposal. What happened to it?
• Describe a time you convinced leadership to invest in something with no precedent.
• What is the biggest opportunity your current company is missing?

A STAR answer that works: Situation: My team maintained 40 internal dashboards, each serving one stakeholder. Task: I was asked to reduce maintenance cost 20 percent. Action: Instead, I proposed killing all 40 and building one self-serve metrics layer, wrote the two-year cost comparison, and recruited two other team leads whose teams had the same disease. We pitched it as an org investment, not a team project. Result: Funded with 6 engineers. Maintenance cost fell 70 percent, not 20, and stakeholder teams built 200+ views without us.

The trap: vision without a first step. Think Big answers need the bridge from big idea to the thing you did in week one.

9. Bias for Action

Speed matters, and most decisions are reversible. The interviewer wants proof you can tell reversible from irreversible and act accordingly.

Questions:

• Tell me about a decision you made in under a day that others wanted to study for weeks.
• Describe a calculated risk that went wrong. What was the calculation?
• When has moving fast hurt you?

A STAR answer that works: Situation: A pricing bug was overcharging a customer segment on a Friday afternoon. Legal review of the fix messaging would take until Tuesday. Task: I owned the response. Action: I classified the refund itself as reversible, issued refunds immediately with a plain one-line notice, and queued the full legal-approved communication for Tuesday. I documented the decision and told my VP the same hour. Result: 1,800 customers refunded before the weekend, 4 support tickets total, zero churn in the affected segment. Waiting would have meant a weekend of visible overcharges.

The trap: speed stories with no risk assessment. Bias for Action is not impulsiveness; the interviewer needs to hear the reversibility check.

10. Frugality

Constraints breed invention: do more with less. This is the most Amazon-specific principle, and unprepared candidates have nothing for it.

Questions:

• Tell me about a time you delivered a result with a fraction of the expected budget.
• Describe a constraint that made your solution better.
• Where have you cut spend that others thought was untouchable?

A STAR answer that works: Situation: Our data pipeline bill grew 40 percent year over year while usage grew 10. Task: Nobody owned the bill; I took it. Action: I traced 60 percent of cost to three abandoned experiment pipelines, built a two-line cost tag per pipeline owner, and set an auto-expiry default for new experiments. Result: $410K annual savings, and the cost-tag report now goes to every team monthly, which changed behavior more than the cleanup did.

The trap: confusing frugality with cheapness. Cutting something that later broke is a negative signal. The story must show intelligence about which costs matter.

11. Earn Trust

Leaders are candid, listen well, and criticize themselves in public. The strongest Earn Trust stories start with you being wrong.

Questions:

• Tell me about a time you rebuilt trust after breaking it.
• Describe delivering bad news to leadership.
• Tell me about criticism you received publicly. How did you respond?

A STAR answer that works: Situation: A migration I championed corrupted records for 3 percent of accounts. Task: I owned both the fix and the communication. Action: I published the incident write-up under my own name, root cause and my own decision errors included, presented it to the affected teams before my chain of command asked, and shipped the two process changes I had promised within a month. Result: The write-up became the team's incident template. Six months later the same teams approved my next migration proposal without escalation, which is what trust looks like operationally.

The trap: stories where you earned trust by being right. Amazon wants candor under bad news, not credibility from good news.

12. Dive Deep

Leaders operate at all altitudes, and no metric is trusted until it survives contact with an anecdote. The interviewer will test whether your depth is real by probing your own story's details.

Questions:

• Tell me about a time a metric looked fine but something felt wrong.
• Describe the deepest you have gone to find a root cause.
• When did digging into details change a decision your team had already made?

A STAR answer that works: Situation: Conversion looked flat quarter over quarter, and the dashboard said all was well. Task: A single seller complaint about checkout latency did not match the dashboard. Action: I pulled raw session data myself instead of delegating, segmented by payment method, and found one provider degrading 8 percent of sessions, invisible in the aggregate. I sat with support for a day to size the complaint pattern. Result: Provider failover shipped in two weeks; conversion in the affected segment rose 6 percent. The aggregate dashboard now ships with segment alarms.

The trap: delegation language. "I asked the team to investigate" is the anti-answer for this principle. The interviewer needs you personally in the data.

13. Have Backbone; Disagree and Commit

Leaders challenge decisions respectfully, then commit fully once decided. Both halves are scored. Most failed answers contain only one.

Questions:

• Tell me about a time you disagreed with your manager and lost. What did you do next?
• Describe pushing back on a decision when it was uncomfortable.
• Have you ever committed to a plan you believed was wrong? Walk me through it.

A STAR answer that works: Situation: My director chose a vendor platform I believed would not scale past a year. Task: I owed the team my real assessment and then a functioning integration either way. Action: I wrote a one-page dissent with data, presented it once, directly, in the decision meeting. The decision went the other way. I then built the integration properly, no slow-rolling, and set up the load tests that would tell us early if I had been right. Result: The platform held for 18 months, longer than I predicted. When we did migrate, my load-test data made the case in one meeting, and my director sponsored my promotion that cycle.

The trap: sabotage disguised as commitment. If your story includes "and then I was proven right because it failed," the interviewer hears that you slow-rolled the decision.

14. Deliver Results

Leaders hit the number despite setbacks. This principle wants your recovery story, not your smooth-quarter story.

Questions:

• Tell me about a commitment you nearly missed. How did you land it?
• Describe delivering under a deadline you thought was impossible.
• Tell me about a time you had to drop something important to protect something more important.

A STAR answer that works: Situation: Six weeks from a contractual deadline, a dependency team lost half its staff to a reorg and our launch date went from green to red. Task: The date was contractual; the penalty was $200K plus the account's trust. Action: I cut scope to the contractual minimum, published the cut list to the customer with the recovery date, and personally took over the riskiest integration piece rather than re-delegating it. Result: Shipped the minimum on the contract date, full scope four weeks later. The customer renewed early, citing the cut-list transparency specifically.

The trap: heroics without prioritization. All-nighter stories score worse than scope-cut stories, because one is stamina and the other is judgment.

15. Strive to be Earth's Best Employer

Leaders build safer, more empowering teams. Added in 2021, this one is asked mostly of managers and senior ICs who shape team environments.

Questions:

• Tell me about a time you made your team measurably better to work on.
• Describe spotting and fixing a burnout risk before it became attrition.
• What have you changed about how your team works that outlasted you?

A STAR answer that works: Situation: My team's on-call was burning people out; two engineers flagged interview processes elsewhere. Task: Attrition risk was mine to fix, without lowering our reliability bar. Action: I cut the pager load in half by fixing the top three alert sources, moved on-call to a follow-the-sun split with a sister team, and made "pages per shift" a metric my director saw monthly. Result: Zero regretted attrition over the next year, pages per shift down 70 percent, and both engineers who had been interviewing stayed and were later promoted.

The trap: perks stories. Team lunches are not this principle. It scores structural changes to how work happens.

16. Success and Scale Bring Broad Responsibility

Amazon's newest principle: your decisions ripple beyond your team, so act like it. Expect it in senior loops, especially for roles with customer, community, or environmental surface area.

Questions:

• Tell me about a time you considered second-order effects of a decision that others ignored.
• Describe a call where the right thing for your metrics was wrong for someone outside your org.
• How have you handled a decision with consequences you could not fully predict?

A STAR answer that works: Situation: We could hit our growth target fastest by defaulting users into data sharing with an opt-out. Task: The target was mine; so was the recommendation. Action: I modeled the trust cost, found opt-out defaults correlated with support complaints and press risk in two comparable launches, and proposed opt-in with an incentive instead, accepting a slower ramp. I put both projections in front of leadership rather than only mine. Result: We hit the target one quarter later than the aggressive path, with complaint rates flat and zero press incidents, while a competitor's opt-out launch that quarter became a news cycle.

The trap: philosophy without a decision. This principle needs a moment where you personally traded a metric for a broader outcome.

Your 2-week prep plan

Days 1 to 4: build the story bank. Write 8 to 10 stories from the last five years. For each, one line per STAR stage, and a number in every Result. Senior candidates should pull from cross-team, cross-org, and customer-facing work, not solo output.

Days 5 to 7: map stories to principles. Each story covers 2 or 3 principles. Check coverage: the three killer principles (Customer Obsession, Ownership, Have Backbone) need your strongest material. Frugality and Earth's Best Employer are where most banks have holes; fill them now.

Days 8 to 14: practice against follow-ups, out loud. Reading your stories is not preparation. You need a partner or an interviewer that probes three levels deep on every answer, because that is what the loop does. Skillora's AI mock interviewer runs full Amazon-style behavioral rounds: it asks Leadership Principle questions, pushes adaptive follow-ups on your actual answers, and scores each response against the principle it maps to. Run two full mocks before your loop, and your first real "tell me about a time" will be your fortieth, not your first.

For the rest of your prep stack, see our reviews of the best interview preparation websites and AI job search tools.