The one interviewer in the loop who is happy to say no
Most people in your interview loop have a reason to want you to pass. The recruiter wants to close the role. The hiring manager wants the empty seat filled. The bar raiser, or whoever runs the values round, is the exception: their reward comes from protecting the quality of every future hire, not from filling this one seat. They are usually trained, and at many companies they sit outside the hiring team entirely, so short-term pressure and the manager's enthusiasm do not move them. You do not pass this round by being likeable or by reciting the company values back. You pass it by giving concrete evidence that you raise the standard around you and that your values hold when they cost you something.
Short answer: Treat the bar-raiser or values round as an evidence test, not a personality test. The interviewer is checking two things: would hiring you lift the team's average rather than dilute it, and do your stated values survive real pressure. Answer both with specific stories where you left something measurably better than you found it and where you held a line when the easy path was to drop it. Own what you did, name what you would do differently, and never turn a disagreement into a story about being right.
What this round actually is, and what it is not
The best-known version is Amazon's Bar Raiser. Carlos Arguelles, a Senior Principal Engineer at Amazon who has served as one, describes the role plainly: a Bar Raiser is a specially trained interviewer in the loop who leads the debrief and has "veto power over all hiring decisions (a hire cannot happen without your thumbs up)." They are typically pulled from a different team, and often a different job family, so they carry no stake in whether this particular seat gets filled today.
Plenty of companies that have never heard the phrase "bar raiser" run the same round under other names: a values interview, a culture-add conversation, a leadership-principles round, or a deceptively casual "meet the team" that is quietly scoring the same things. The label varies. The job of the person across the table does not: they are the check against a team talking itself into a hire because it is tired of interviewing.
That is why the round exists at all. When the people who need the seat filled also decide who fills it, standards drift downward under deadline pressure and affinity bias creeps in. A separate, trained evaluator is the counterweight.
The bar raiser is not there to like you. They are there to answer one question the hiring manager cannot ask themselves honestly: is this person better than the people we already have?
What "raise the bar" actually means
The phrase gets repeated so often it stops meaning anything, so anchor it to the source. Amazon's own Leadership Principles page defines the "Hire and Develop the Best" principle as: "Leaders raise the performance bar with every hire and promotion." Read literally, that is a mathematical claim. Every hire should pull the average up, which means roughly half of a healthy team's existing members would not clear the bar today. Your job in this round is to give the interviewer enough evidence to believe you are in the top half of the people they would otherwise hire.
The trap is that most candidates answer values questions with claims instead of evidence. "I really care about code quality" is a claim. Anyone can say it. The bar raiser is trained to ignore the claim and dig for the proof underneath it, which is why so many answers that sound good collapse on the second follow-up. Here is the difference between a claim and the evidence that survives probing.
| The value you might claim | The evidence that proves it | What makes it hollow |
|---|---|---|
| I hold a high quality bar | You changed a default so the next person could not skip a step you thought mattered | Naming a time you personally wrote careful code once |
| I take ownership | You fixed a problem that was technically not yours and no one asked you to | Owning something you were already assigned |
| I have backbone | You disagreed with someone more senior, lost some of it, and can say what you learned | A story where you were simply right and they were wrong |
| I raise the people around me | Someone got measurably better or unblocked because of a thing you built or taught | Being generally friendly and helpful |
| I use good judgement under ambiguity | You made a reversible bet with incomplete data and named the risk out loud | A decision that was obvious in hindsight |
The pattern across the strong column: every one is a change that outlived the moment. You did not just do good work, you left the environment better than you found it, so the next person inherits a higher floor. That is what raising the bar looks like in practice, and it is exactly what the interviewer is fishing for.
A bar-raiser deep-dive, taken apart
The core of this round is one story followed to the bottom. The interviewer picks a thread and pulls until they hit either bedrock or air. Here is a real-shaped exchange between Wei, a staff engineer from a different org acting as bar raiser, and Kiran, a platform engineer interviewing for a senior role. Watch what Wei is actually testing at each turn.
Wei: Tell me about a time you improved something no one asked you to improve.
Kiran: Our release pipeline had a required test gate, but the tests had gotten flaky. When something failed, the habit on the team was to just re-run the pipeline until it went green. Under deadline pressure that had quietly become the norm.
Wei: What was wrong with re-running until green?
Kiran: It meant the gate was not really a gate. A genuine regression that failed once and passed on a retry would ship. We shipped two that quarter that way. The gate looked green in the logs, but it was not protecting anything.
Wei: So what did you do? Not the team. You.
Kiran: I built a quarantine step. If a test failed then passed on retry across a rolling window, it got auto-flagged as flaky, moved to a separate non-blocking lane, and assigned an owner with a two-week deadline to fix or delete it. Then I made the main gate required-green with no manual override, because the override was the actual hole.
Wei: Removing the override sounds like it would have blocked a release. Did anyone push back?
Kiran: Yes. A senior engineer wanted to disable the gate entirely to hit a launch date. I disagreed. I did not win the room by insisting; I proposed the quarantine as the middle path, because it kept the launch date and closed the hole at the same time. He still thought it was overkill, but he agreed to try it for one release.
Wei: And if it had blocked that launch?
Kiran: Then I would have been wrong in front of someone senior, on a visible date. I accepted that risk because a gate you can always bypass is worse than no gate, since it lies to you. It did not block the launch, and the quarantine lane became the default for three other teams within a quarter.
Wei: What would you do differently?
Kiran: I shipped the required-green change a bit too abruptly. I should have socialized the quarantine idea a week earlier so the senior engineer was not hearing it for the first time in a launch meeting. The idea was right; the rollout created more friction than it needed to.
Notice how little of this is about the code. Wei never asked how the quarantine was implemented. Every question was aimed at a value: did you own it beyond your remit, did you hold a line under pressure, did you handle disagreement without needing to be right, did the change outlive the moment, and can you name your own mistake without being asked twice. The last answer, the "what would you do differently," is the one that most often separates a strong rating from a lukewarm one, because it proves the self-awareness the rest of the story only implies.
The scorecard the bar raiser is filling
While you talk, the interviewer is mapping your answers onto a rubric, not a gut feeling. Structured interviewing is deliberate here: Google's re:Work guide notes that "we use the same interview questions, grade candidate responses on an identical scale, and make hiring decisions based on consistent, predetermined qualifications," and that this "result[s] in increased predictive validity and decreased differences between demographic groups." The bar raiser round is the most structured seat in the loop precisely because it is the one guarding against bias. Knowing the rough shape of the scorecard tells you what to supply.
| Signal | What earns a strong mark | What earns a no |
|---|---|---|
| Ownership | Fixed something outside your lane, unprompted | Only delivered what you were assigned |
| Raising the bar | Left a durable change others now inherit | A one-off good result with no lasting effect |
| Backbone | Disagreed up, adapted, and learned | Avoided conflict, or won by being stubborn |
| Judgement | Named the risk and the tradeoff out loud | Presented the decision as obvious |
| Self-awareness | Volunteered a real mistake and the lesson | Could not name anything you would change |
| Data, not slogans | Concrete numbers, names, and outcomes | Values recited back as adjectives |
The veto matters here. As the Bar Raiser account describes, a hire cannot happen without their sign-off, which means a strong performance in every technical round can still be overridden by a weak values round. Treat this seat as a gate that is fully independent of how the coding rounds went, because for the person filling it, that is exactly what it is.
Where values answers quietly fail
The failures in this round are rarely dramatic. They are small tells that a trained interviewer is specifically listening for.
- Slogans instead of stories. Answering "how do you handle conflict" with a philosophy of conflict rather than one time you actually had one. The rubric has no box for philosophy.
- The "we" that hides the "I". A story where everything is "we decided" and "we shipped" so the interviewer cannot tell what you personally did. Expect the "what did you do" follow-up, and answer it in the first person.
- The hero with no team. The opposite failure: a story where you single-handedly saved everything and no one else contributed. It reads as either exaggeration or someone who does not raise the people around them.
- Disagreement as being right. Backbone stories that are really about how you were correct and the other person was foolish. The signal the interviewer wants is that you can hold a position, lose part of it, and commit anyway.
- No mistake you can name. Freezing on "what would you do differently" or offering a fake weakness. Not having a real answer reads as either low self-awareness or a rehearsed candidate, and both lower the score.
- Values that clash with the actual principles. If a company publishes its values, and many do, an answer that celebrates the opposite of one lands badly. Read them first and pick stories that align without pandering.
When the round wears a different name
The seat is consistent worldwide, but its framing shifts, and the shifts matter.
Culture fit versus culture add. Older versions of this round asked whether you "fit," which BetterUp warns "can lead to homogeneous teams, missing out on unique perspectives" and "can reinforce existing biases, even unconsciously." Many companies have deliberately moved to culture add, defined as "how a candidate can enhance and diversify an organization's existing culture, rather than simply fitting into it." Practically, this means you should not sand off what makes your background different. The thing your last three teams did not have is often the strongest answer to why you raise the bar, not a liability to hide.
Region and register. How directly you are expected to describe a disagreement varies by workplace culture. In some regions, naming that you overrode a senior colleague reads as confidence; in others, it reads as a red flag, and the same story lands better framed as building consensus. The underlying signal, that you held a standard under pressure, is universal. The volume you deliver it at is not, so read the room and adjust the framing rather than the substance.
The culture-fit-as-bias trap, pointed the other way. If the round feels like the interviewer is screening for whether you would fit in socially rather than whether you would raise the bar, that is a signal about the company, not about you. A values round should be structured and evidence-based. One that runs on vibes is a place where, as the research on unstructured culture-fit interviews shows, decisions "frequently reflect the interviewer's personal preferences." That is worth noticing while you are also being evaluated.
FAQ
Is the bar raiser only an Amazon thing? No. The trained-interviewer-with-veto version is most associated with Amazon, but the round it represents, an independent evaluator checking values and standards, exists at many companies under names like values interview, culture-add round, or leadership-principles interview. The preparation is the same wherever it appears.
How do I prepare stories for a round when I do not know the questions? Build a small bank of three to five stories, each of which can be told to show ownership, backbone, a durable improvement, and a real mistake. One strong, layered story usually covers several values because a good bar raiser will pull the same story in multiple directions rather than asking a fresh question each time.
Should I read the company's published values before the round? Yes, and pick stories that genuinely align with them, but never recite the values back. The interviewer is trained to spot pandering. Alignment should show through your examples, not through you naming the principle.
What if I genuinely disagree with one of the company's values? It is fair to answer honestly and even to name a tension, as long as you do it thoughtfully. A candidate who can respectfully push on a value often scores better than one who agrees with everything, because agreeing with everything reads as not having thought about it.
How much detail should the mistake in my "what would you do differently" answer have? Enough to be real: a specific thing you would change and why, tied to the actual story. Avoid the fake weakness ("I care too much"). A concrete, low-drama mistake with a clear lesson is the strongest possible answer to that question.
Can a great technical loop still fail on this round? Yes. Where the round carries a veto, a weak values interview can override strong technical feedback entirely. Prepare for it with the same seriousness you give a coding round, because for the person filling this seat, it is decisive.
What if I do not have a big, impressive story? Scale is not the signal; a change that outlived the moment is. Improving a small process your five-person team still uses is stronger evidence of raising the bar than a large project where nothing you did persisted.
Sources
- Carlos Arguelles, "Memoirs of an Amazon Bar Raiser," Medium (a Senior Principal Engineer's first-hand account of the role, its training, and its veto power): https://medium.com/geekculture/memoirs-of-an-amazon-bar-raiser-718e36241310
- Amazon, "Leadership Principles" (the "Hire and Develop the Best" and "Insist on the Highest Standards" definitions): https://www.aboutamazon.com/about-us/leadership-principles
- Google re:Work, "A guide to structured interviewing for better hiring practices" (same questions, identical scale, reduced bias): https://rework.withgoogle.com/intl/en/guides/a-guide-to-structured-interviewing-for-better-hiring-practices
- BetterUp, "Culture fit vs. culture add" (definitions and the homogeneity and bias risks of hiring for fit alone): https://www.betterup.com/blog/cultural-fit
Where to take this next
- Build the raw material this round runs on with behavioural interview frameworks and the STAR framework for behavioral interviews.
- See how this seat differs from the manager who owns the role in the hiring manager interview round.
- Prepare for the format where a bar raiser often sits alongside others in how to handle a panel interview.