Amazon Bar Raiser: They Hold Veto Power. Here's What They Want.

You've cleared the phone screen. You've made it through three or four solid rounds where every interviewer said you're great. Then one more interviewer shows up. They're not on the team you're joining, they have no stake in whether the role gets filled, and they can block your hire even if every other person in the room votes yes.

That's the Amazon Bar Raiser. And most candidates walk in without knowing what they're actually evaluating.

The Bar Raiser isn't there to test your code or quiz your system design skills. They're there to answer one question: does this person raise Amazon's bar? Not meet it. Raise it. If you're merely adequate, you don't pass. This is not a formality. It is the most consequential conversation in your loop, and you won't know which interviewer it is until after you leave.

What the Amazon Bar Raiser Actually Is

Amazon launched the program in 1999, when then-CIO Rick Dalzell drew inspiration from Microsoft's "as appropriate" final interviewer. Microsoft's version was hit-or-miss. Sometimes it happened, sometimes it didn't. Dalzell formalized it and made it mandatory on every loop.

The original name was the "Barkeeper Program," which made it sound like maintaining existing standards. They changed the name immediately. The point was never to hold the line. It was to raise it with every single hire.

Today there are more than 10,000 Bar Raisers and Bar Raisers-in-Training globally. They're not full-time hiring people. They're software engineers, product managers, marketing leads, and operations folks who volunteer their time and get certified after three months to a year of training and shadowing.

The structural detail that matters most: the Bar Raiser comes from a different organization than the team you're interviewing with, and has no business interest in filling this role. A hiring manager under pressure to staff up a team wants to say yes. The Bar Raiser doesn't care either way. That asymmetry is the entire design.

The 50 Percent Rule

Every new hire must be better than at least 50 percent of the people currently doing that job at Amazon.

Not "good." Not "a solid engineer." Better than half the existing population in that role.

This is where most candidates trip. They prepare to demonstrate competence. But the Bar Raiser is calibrating against a population, not an absolute threshold. An answer that's technically correct but shows no genuine ownership, no bias toward action, and no evidence of thinking beyond the immediate problem will read as fine to a hiring manager who's been trying to fill a seat for two months. The Bar Raiser will vote no.

The question they're actually asking with every story you tell is: would this person, already on our team, be in the top half?

What They Assess

The Bar Raiser interviews like every other interviewer on the loop, which means behavioral questions mapped to Amazon's 16 Leadership Principles. In a five-person loop, each interviewer typically covers two or three LPs. The Bar Raiser might also include a functional component: a coding problem for an SDE role, a product scenario for a PM.

Their primary lens is behavioral, though. They'll pick one or two LPs and go three to five questions deep on each. Not three questions total. Three to five follow-up probes per story. They're not collecting stories. They're stress-testing the ones you give them.

A few principles show up disproportionately often because they're genuinely hard to fake at depth:

Ownership. Not "we deployed the fix." What did you personally do when the system failed on a Sunday?
Dive Deep. Not "I analyzed the data." What specific number did you find that no one else was looking at, and what did you do with it?
Have Backbone; Disagree and Commit. Not "I spoke up." Did you actually push back on your manager with data, and then commit fully when the decision went the other way?

The Bar Raiser is looking for the gap between what you say you did and what actually happened. They'll ask "what did your manager think?" and "what would you have done differently?" to find it.

Leonardo DiCaprio laughing meme: SOFTWARE ENGINEERS ARE SOME OF THE BEST ACTORS. OUR STAGE IS THE BEHAVIORAL INTERVIEW ROUND.

Every engineer going into round four already performing their "genuine enthusiasm for ownership" arc.

The Debrief Is Where Votes Change

After your loop ends, every interviewer submits written feedback independently, before talking to anyone else. Then comes a structured debrief meeting.

The Bar Raiser facilitates it. Not the hiring manager.

The debrief starts with everyone reading all the written assessments in silence for 10 to 15 minutes before anyone speaks. Each interviewer only gathered behavioral data on their assigned LPs, roughly one-fifth of the total picture in a five-person loop. Reading the complete set changes votes. It happens frequently. Someone who voted hire in their own feedback quietly updates their view after reading someone else's notes.

Once discussion opens, the Bar Raiser uses a Socratic approach. They don't declare a position first. They ask questions, push interviewers to justify their assessments with specific behavioral evidence, and call out conclusions that aren't grounded in what actually happened in the interview. If an interviewer says "they seemed sharp," the Bar Raiser asks them to point to the specific behavior that supports that.

The program was designed to counter three specific biases: personal affinity (hiring someone because they remind you of yourself), urgency bias (hiring someone adequate because you've been searching for two months), and confirmation bias (letting early opinions color later ones).

The Bar Raiser has veto power. In practice, long-tenured Bar Raisers describe it as a nuclear option that almost never gets used. One veteran reportedly conducted over 700 interviews without once formally overriding a hiring manager. The real mechanism is facilitation, not dictation. But the authority is real, and it changes the dynamic in the room.

This matters for you: your stories don't just stay with the one person who interviewed you. They travel to the whole panel in written form. Write feedback that survives that reading.

The Prep Playbook

You need stories. Specific ones, mapped to LPs, with real numbers.

The prep mistake most candidates make is starting from the LP list. They read "Customer Obsession," think of something vaguely customer-related, and write a rough outline. That produces 16 shallow stories you can't defend under follow-up probes.

Start from your projects. List the 10 to 12 most significant things you've worked on in the past three to five years. For each one, write out the situation, your specific role, four to six concrete actions you took, and two or three quantified results. Then map each project to the LPs it demonstrates. A single rich project might cover Customer Obsession, Ownership, Bias for Action, Dive Deep, and Deliver Results. You're not memorizing 16 different stories. You're building a few deep ones that each stretch across multiple principles.

Senior candidates typically prepare 30 or more stories because in a full on-site loop, you might face 25 to 35 behavioral questions and you cannot reuse a story within the same loop. Interviewers compare notes. If the Bar Raiser sees the same story used for both "Dive Deep" and "Deliver Results," they'll flag it in the debrief.

At minimum, 12 to 15 fully mapped stories before you walk in. For L5 and above, aim for 30. Yes, 30. That's not a typo.

The Follow-Up Is Where You Pass or Fail

Writing stories isn't the hard part. Defending them is.

Practice giving a story, then imagine someone asking "what was YOUR specific contribution to that?" Again. And again. And "what would you have done differently?" And "what was the exact metric before and after?" Candidates who fail the Bar Raiser round typically don't fail on the first telling. They fail on the third follow-up when the precision runs out.

A few patterns that end interviews:

Vague outcomes. "The project went well" is not a result. "User complaints about that flow dropped by 40 percent in the next quarter" is. If you don't have a number, get as close as you can. "Revenue impact was hard to isolate but we reduced churn in that segment by approximately 15 percent" is better than nothing.

Team credit. Amazon's LPs use "leaders" as the subject for a reason. When you say "we," the Bar Raiser will ask what you specifically did. If the honest answer is "I was one of eight engineers who built it," they'll probe until they find something you personally owned or decided. Find that thing first. Lead with it.

No learning from failure. Bar Raisers actively want to hear about things that went wrong. The framing that works: here's what I tried, here's what broke, here's exactly what I learned, here's the change I made next time. Candidates who turn every story into a success narrative come across as either dishonest or lacking self-awareness. Both read as red flags. The broader interview mistakes that get you rejected map almost exactly to Bar Raiser failure modes.

Raccoon sitting at a table looking uncomfortable. Text: Interviewer: so tell me a little about yourself. Me: I'd rather not, I kinda need this job

The correct internal monologue. The Bar Raiser will probe until you share anyway, so you might as well have a good answer ready.

Your Stories Need to Be Real, Not Just Polished

You'll find advice that says "just be yourself" in the Bar Raiser round. That's incomplete.

Being authentic in a behavioral interview means telling real stories from your actual experience, not polished versions of what you think they want to hear. Bar Raisers are trained to detect scripted answers because scripted answers collapse under follow-up. If you're five layers deep into a story and still have specific details to give, that's authentic. If you start hedging on the second follow-up, they'll notice.

The stories need to be yours. The structure needs to be STAR. The metrics need to be real. Those three things together are what "prepared" actually looks like.

How you talk through your thinking under pressure is itself a data point. If you've read about how communication signals competence in technical interviews, the same logic applies here, just with behavioral stories instead of code.

If you want to practice this under real pressure, SpaceComplexity runs voice-based mock interviews with rubric-driven feedback. Running your STAR stories out loud is different from writing them down, and the Bar Raiser round is entirely a spoken exercise.

Amazon Invented This, But It's Not Alone

Amazon's program is the most formalized version of this idea. Microsoft uses an "as appropriate" interviewer for a subset of candidates. Google routes final hiring decisions through a committee review, removing the hiring manager from the equation entirely. Meta has similar independent reviewers for certain roles.

The through-line is the same in each case: someone with no stake in saying yes is the last checkpoint before an offer goes out. If you're prepping for any big tech loop, assume this mechanism exists in some form.

The Short Version

The Bar Raiser is from a different org, has no urgency to fill the role, and can block your hire even if everyone else says yes.
The mandate: you need to be better than 50 percent of current Amazon employees in that role.
They assess LP alignment through behavioral questions, probing three to five layers deep per story.
The debrief happens after you leave. Your stories travel to everyone. Write feedback that survives that reading.
Build your prep from projects, not LP lists. Aim for 12 to 15 stories minimum, 30 for senior roles.
Vague outcomes and team credit are the two fastest ways to a no-hire.
You can't identify the Bar Raiser during the interview. Treat every interviewer like they have veto power.