Amazon Behavioral Interview Questions: The STAR + Leadership Principles Guide

1. The STAR method (done properly)

STAR — Situation, Task, Action, Result — is the story shape Amazon interviewers explicitly want. The mistake most candidates make is spending 80% of their answer on Situation and Task, then rushing the Action and skipping the Result. Bar raisers are grading Action and Result. That's where your judgment shows up.

• Situation (10–15%) — one sentence of context. Team, product, timeline.
• Task (10–15%) — what you specifically owned. Not the team, you.
• Action (50–60%) — what you did, in first person. "I" not "we". Include trade-offs, what you considered and rejected, and who pushed back.
• Result (20%) — measurable outcome. Numbers, percentages, dollars, users, latency. Then a one-line reflection on what you'd do differently.

2. The 16 Amazon Leadership Principles

Amazon publishes 16 Leadership Principles. Interviewers pick 2–3 per round and probe them with behavioral questions. You should have at least two prepared stories per principle — one recent, one senior — so you're never forced to reuse an answer.

3. The questions you'll actually get

Amazon recycles a fairly stable question bank. If you can answer these ten cleanly, you can answer most of what a loop will throw at you:

1. Tell me about a time you had to make a decision with incomplete information.
2. Tell me about a time you failed. What did you learn?
3. Describe a time you disagreed with your manager.
4. Tell me about the most complex problem you've solved.
5. Describe a time you had to push back on a customer.
6. Tell me about a time you didn't hit a deadline.
7. Describe a time you had to work with a difficult teammate.
8. Tell me about a time you had to influence without authority.
9. Tell me about the most innovative thing you've built.
10. Tell me about a time you were on a team that was underperforming.

4. Two full example answers

Ownership — "Tell me about a time you took on something outside your remit."

Situation: Our payments team owned checkout, but customer support was drowning in refund tickets during Q4. Nobody owned the refund UX.

Task: As the checkout tech lead, I decided the refund flow was ours by proximity, even though nobody had assigned it.

Action: I pulled two weeks of support tickets, clustered the top three failure modes, and shipped a self-serve refund page in 11 days. I got pushback from my manager that this was scope creep. I committed to keeping our roadmap dates intact and personally covered the extra hours.

Result: Refund tickets dropped 62% in the first month, saving support ~30 hours a week. My manager later formalised the flow as part of our team's charter. If I did it again, I'd have gotten written air-cover from support leadership first instead of assuming goodwill.

Have Backbone — "Describe a time you disagreed with your manager."

Situation: My manager wanted to launch a pricing change without an A/B test because leadership was pushing for speed.

Task: I had modeled a plausible 8–12% revenue downside and thought we needed at least a two-week holdout.

Action: I wrote a one-page memo with the revenue model, three prior pricing launches at the company, and a proposal for a 10% holdout that added only three days. I walked my manager through it 1:1 before escalating. He disagreed but agreed to bring it to the director. The director sided with me and we ran the holdout.

Result: The holdout caught a 9% revenue regression from an unrelated bug in the new pricing page — we would have shipped a real loss. We rolled back, fixed it, and the eventual launch was flat-to-positive. My manager later said the memo format was what changed his mind and asked me to teach it to the team.

5. The mistakes that fail candidates

• "We" instead of "I". Amazon grades individual contribution. If you can't say "I", the interviewer can't score you.
• No numbers in the Result. "It went well" is unrateable. Put a percentage, a dollar figure, or a latency number.
• One story per principle. Two interviewers may probe the same principle. Reusing an answer signals shallow experience.
• No trade-off. Every real decision has a cost. If your story has no downside, the bar raiser assumes you skipped it.
• Not answering the actual question. When asked about failure, candidates often tell a success story. Interviewers write this down as "did not answer".

6. Practising under real pressure

Reading this guide won't get you the offer. What gets you the offer is saying these stories out loud, on a clock, with someone probing for the weak parts. That's exactly what RoleMatch does — it reads your CV and the Amazon JD, generates a targeted set of behavioral questions mapped to the Leadership Principles, and runs a live AI-voice interview that pushes back when your answer is thin.

After the interview you get a RoleMatch scorecard — a per-answer score, a rating on STAR structure, filler words, speaking pace, and a shareable summary. Most candidates find their first three attempts are 40+ points below their fifth.

Related reading: How RoleMatch scorecards work.