Amazon PM interview process: every stage, the Bar Raiser, and what clears the bar in 2026

The Amazon PM process is the most behaviorally engineered interview loop in big tech. Each interviewer enters your session with 1-3 Leadership Principles pre-assigned, asks one question per principle, and submits an independent vote before any group discussion. The Bar Raiser is embedded in the onsite loop as one of five interviewers, you will not know which one, and their job in the debrief is to answer a specific question: is this candidate better than the median current Amazon employee at this level in this role? That is the raise-the-bar threshold, not a vibe check and not a rubric.

The five stages

Recruiter screen (30 minutes). Background fit and motivation. The recruiter calibrates seniority and flags LP mismatches before anything reaches the hiring manager. If your resume mentions a major project without quantified outcomes, expect a probe here framed as Ownership or Dive Deep.

HM screen (60 minutes). Split roughly 50/50: the first half covers LP questions anchored to your product experience, the second half covers product and analytical framing. Customer Obsession and Ownership are the highest-priority gaps the HM is checking for. Do not treat this as a warmup.

Written assessment. Sent after the HM screen, submitted before the onsite loop. Format: a 1-2 page prose memo, not a slide deck. The prompt is usually LP-based, such as a high-stakes decision made with incomplete data, or a product direction you disagreed with. Key fact most guides miss: interviewers read the memo for LP alignment even when the prompt does not name a Leadership Principle explicitly. Every paragraph should be traceable to a principle. Write in Amazon’s internal doc style: direct, specific, willing to name what you would do differently. The two failure modes are vague narratives without outcome metrics, and bullet-formatted responses where Amazon expects prose reasoning.

Onsite loop (5 x 55-minute interviews). Five sessions back to back. Interviewers are pre-assigned their LP blocks before the day begins. No two interviewers probe the same LP. Across five rounds covering 1-3 LPs each, the loop is designed to surface all 16 principles. You will not know which interviewer is the Bar Raiser.

Debrief. Interviewers submit independent votes and written justifications before the group convenes, an explicit anti-anchoring design. The Bar Raiser leads the debrief, reads every independent assessment, and can veto any hire outcome. Amazon does not overrule the Bar Raiser.

The Bar Raiser: what the veto actually means

The Bar Raiser is an L6+ Amazon employee from a different organization than the hiring team. They are trained over 6-12 months specifically for this role. Amazon does not disclose which interviewer they are before or during the loop.

Most guides describe the Bar Raiser as holding an absolute unilateral veto. That is partly right and partly myth. The more accurate picture, from people who have been through the debrief process: the Bar Raiser moderates the debrief, anchors calibration to cross-Amazon standards, and can block a hire. Hiring managers can accept documented risk on minor concerns. True unilateral blocks where the Bar Raiser fires against unanimous hire votes from all other interviewers do happen, but they are not the common case. The more common outcome is that a strong Bar Raiser objection reshapes the group vote.

The Bar Raiser’s calibration question is specific: is this candidate better than the median current Amazon employee at this level in this role? Not “is this person good?” Not “can they do the job?” Better than median, relative to a real population. That standard means a competent candidate who lands at median across all LPs fails, even if no single answer was wrong.

The typical veto is not a wrong answer. It is an answer that sounds confident but cannot survive the third follow-up question. If you said retention increased 20%, the Bar Raiser asks: what was the measurement window? What was the denominator? What changed in the product specifically? Candidates who prepared polished answers but did not deeply own the work they described get caught here.

Common triggers by LP:

• Dive Deep: You cite a metric but cannot explain who measured it, what the cohort definition was, or what changed in a later period.
• Customer Obsession: Your story grounds customer insight in internal stakeholder preference or an NPS score rather than direct research or behavioral data.
• Ownership: The decision was made by committee. Your role was coordination, not the call.
• Bias for Action: You frame a fast decision, but the timeline reveals you waited for alignment before moving.
• Are Right A Lot: You defend a position with logic but cannot name the data that would change your mind.
• Have Backbone: Disagree and Commit: You describe disagreeing, but your story ends with the other person being wrong rather than you committing to execute despite the disagreement.

LP pre-assignment: how coverage actually works

The loop covers all 16 LPs by distributing them across five interviewers, 1-3 per interviewer, before the loop begins. This means two things in practice.

First, no two interviewers will ask about the same LP. The redundancy candidates dread (“will I get asked Ownership twice?”) does not happen by design. Second, the LP you get in round one determines which LP another interviewer gets in round four. The distribution is coordinated, not random.

The five LPs that appear in almost every PM loop across all levels: Customer Obsession, Ownership, Dive Deep, Deliver Results, Bias for Action. Prepare three stories per principle with deep Dive Deep layers on each.

For Senior PM and PM-T: add Are Right A Lot, Invent and Simplify, Have Backbone: Disagree and Commit. These carry more weight at L6 and above.

LPs 15 and 16 (Strive to be Earth’s Best Employer and Success and Scale Bring Broad Responsibility, added in 2021) appear less often in PM loops but surface in Staff-equivalent conversations, usually framed around decisions with broad stakeholder or societal impact.

AI fluency in 2026

The LP “Invent and Simplify” has developed new meaning at Amazon. With AI making almost any feature technically feasible, interviewers are now using this principle to probe whether candidates can identify which problems are worth building for and can cut ruthlessly to the lovable core rather than shipping bloated AI wrappers.

A weak Invent and Simplify answer in 2026 describes an AI feature you added to a product. A strong one explains how you used AI to remove a step the customer should not have had to take, and then names the features that became redundant and were killed. That is the 2026 version of simplify. Abstract opinions about AI transformation do not pass. Interviewers expect a specific product example with a specific AI mechanism and a measurable outcome.

“Dive Deep” has new teeth in technical loops. Interviewers increasingly probe whether candidates understand the economics of AI in production: cost per query, latency versus accuracy tradeoffs, how you designed evals to detect when the model was wrong. PMs who cannot speak to model behavior in production are falling below bar in PM-T loops specifically.

PM-T: what actually differs

PM-T (Product Manager Technical) is a separate track used primarily for infrastructure, device, and Alexa-domain roles. Not a seniority step above PM. The structural differences:

• An additional technical phone screen (60 minutes) sits between the HM screen and the written assessment.
• The written exercise targets a complex technical project specifically, not a general LP memo.
• One dedicated technical depth round runs inside the onsite loop.
• No coding is required. What is required is system design fluency: the ability to reason about scalability tradeoffs, service boundaries, and data flow without being handed a whiteboard problem with an expected algorithmic answer. Think quasi-system design (tradeoffs, architecture choices, scalability reasoning) rather than leetcode.

If the role is not in an infra, device, or Alexa domain, applying to the standard PM track is usually correct. PM-T L6 total compensation ranges roughly $285K-$385K in 2026; L7 ranges $357K-$798K; L8 (Director) ranges $775K-$1.065M. Full breakdown at Amazon PM salary by level.

Amazon’s level mapping: L5 is PM, L6 is Senior PM, L7 is Principal PM, L8 is Director. L9 does not exist at Amazon.

How to build a story bank that clears the bar

Fourteen to sixteen stories, each mapped to a primary LP and one secondary LP for pivots when an interviewer asks from a different angle. For each story: one sentence on the context, one on the specific complication you personally owned, two or three on the specific actions you took in first person (not “we”), and the measurable outcome with your honest role in producing it.

Prepare the third follow-up before you need it. Amazon interviewers, especially the Bar Raiser, are calibrated to probe past the first answer into the decision quality underneath. If you cannot explain how you measured the result, what you would change with hindsight, or what information would have caused you to decide differently, the story is not ready.

The STAR framework handles the shape. The product round uses Working Backwards. For LP question practice by question type, see Amazon Leadership Principles interview.