Why Meta's Loop Is Different

Meta's behavioural loop is shaped by a culture that explicitly values velocity, direct disagreement, and outsized impact. The company has revised its values list multiple times; the current published values are:

1. Move Fast (the original cultural anchor; revised from 'Move fast and break things' but the velocity expectation remains).
2. Focus on Long-Term Impact (added to balance Move Fast: speed in service of decade-scale outcomes, not quarterly wins).
3. Build Awesome Things (craftsmanship, ambition, and pride in the artefact).
4. Live in the Future (working on the company you want to be, not the company you are; tied to AI and AR/VR investments specifically).
5. Be Direct and Respect Your Colleagues (direct feedback as a core operating mechanism, paired with an explicit respect commitment).
6. Meta, Metamates, Meta (a hierarchy of priorities: company first, teammates second, individual third; pronounced 'Meta-Mates-Meta').

In practice the loop is graded on a smaller subset of these. The behavioural round at Meta, often called the 'People' round, is explicitly looking for evidence that the candidate can:

• Move fast in practice (real velocity stories, not just velocity-flavoured ones).
• Disagree directly (with managers, peers, and senior stakeholders, productively).
• Take outsized ownership (cross-team scope, multi-quarter impact).
• Receive feedback non-defensively (since 'be direct' cuts both ways).
• Operate in the future-state (work on what is hard and ambitious, not what is comfortable).

The distinctive structural feature of Meta's loop is twofold:

1. The People round is grade-determinative. Unlike Google's Googleyness round, which is one of several signals, Meta's People round is treated as one of the strongest hire-signal rounds in the loop. A weak People round is hard to overcome.
2. Internal shorthand for archetypes. Meta interviewers, in debriefs, often categorise candidates as 'Jedi' (craftsmanship-led), 'Pirate' (bias to ship), or 'Ninja' (cross-team scope and influence). This shorthand is internal, but understanding that interviewers think in these archetypes helps you tell stories that map cleanly to one or more of them.

The Six Values in Practice

Not all six values carry equal weight in the People round. The frequency with which they show up:

• Be Direct and Respect Your Colleagues (almost every People round, often the opening probe).
• Move Fast (almost every People round, especially for IC and senior IC roles).
• Focus on Long-Term Impact (most senior People rounds, especially for E5+).
• Meta, Metamates, Meta (most People rounds, often probed via team or company-prioritisation tradeoffs).
• Build Awesome Things (most engineering loops, often woven into the technical and system design rounds).
• Live in the Future (more common at E5+, and increasingly common across all levels post-2023 reorientation around AI).

The practical implication: bank 6 to 8 stories that collectively cover the top four values, with Move Fast and Be Direct having two stories each because they are probed most frequently.

A Note on 'Be Direct'

Meta's most distinctive cultural value, and the one most often misread by candidates, is Be Direct. The misreading is to interpret it as 'speak bluntly without considering the listener'. The actual value, in practice, is paired: be direct AND respect your colleagues. Strong stories show both halves:

• The candidate said something hard that needed to be said.
• The candidate said it in a way that the recipient could hear it and act on it.
• The relationship was intact or improved afterwards.

Candidates who skip the second and third beats fail this value badly. The interviewer reads them as 'this person mistakes bluntness for directness', which is a strong red flag at Meta specifically.

How the Loop Works (Format)

A typical Meta onsite for an E4 software engineer:

• 5 rounds of 45 minutes
• 2 coding rounds (medium to hard)
• 1 system design round (for E5+; E4 may have a third coding round instead)
• 1 People round (the explicit behavioural round)
• 1 design or domain round (varies by role)

For E5 and above:

• 2 coding rounds
• 2 system design rounds (one product-flavoured, one infra-flavoured)
• 1 People round (slightly longer at senior levels, sometimes 60 minutes)

The People Round

The People round at Meta is structured but not scripted. The interviewer typically asks 4 to 6 prompts, each grounded in one of the values. They are looking for:

• Real stories with real specifics. Names of teams, dates, sizes, numbers.
• A point of view. Meta interviewers grade for whether the candidate has formed opinions about how to work, not just executed against instructions.
• A directness signal. Often the People round will include a 'tell me about a time you disagreed' question, and the interviewer is grading the candidate's willingness to recount the disagreement honestly versus softening it.
• A velocity signal. Stories where the candidate explicitly traded scope or quality for speed (with reasoning) score better than stories where the candidate optimised for thoroughness without naming the velocity tradeoff.

The interviewer in the People round is usually a senior IC or a manager, not a recruiter. They are calibrated against the team's bar and they generally write a clear 'hire' or 'no hire' signal in the debrief.

Value-to-Question Mapping

Model Answers Tailored to Meta

Worked Example 1: The Same Story, Reframed for Two Values

The underlying story is a feed-ranking experiment at a content platform.

Underlying story: As an E4 engineer at a content platform, I ran an experiment to ship a new feed-ranking model. The model offline metrics looked good. We had two paths: a thorough 6-week multi-stage rollout with deep evaluation, or a fast 2-week single-stage launch with light evaluation and aggressive monitoring. I argued for the fast path with strong rollback gates. We shipped in 2 weeks, hit a regression on a niche user segment in week 3, rolled back partially within hours, fixed the issue, and relaunched. The final outcome was a 3.2% lift in core engagement, captured 5 weeks earlier than the slow path would have allowed.

Framing 1: Move Fast

'I want to share a time I argued for shipping faster than the team's default and what happened. At my previous company I was an E4 engineer working on feed ranking. We had a new model that looked promising in offline evaluation. The team's default playbook was a 6-week multi-stage rollout: small population, medium population, full population, with a week of evaluation between each stage. I proposed a 2-week single-stage rollout instead, with aggressive rollback gates: automated alerts on engagement, NPS, and content-mix metrics, and a one-click rollback if any tripped.

The disagreement was real. Our tech lead worried about a regression on power users. My read was that the rollback gates would catch any regression within hours, and that the 4 weeks of additional rollout time was 4 weeks of foregone engagement gain if the model was actually good. I wrote a one-page argument with the rollback design specified, presented to the team, and committed to writing the alerting myself if approved. They approved it.

We shipped in week 2. In week 3 the alerts tripped: a niche user segment (heavy creators) was seeing a regression in their reach. We rolled back the model for that segment within 90 minutes while keeping it live for everyone else. I spent three days isolating the cause (a feature interaction with creator-tier signals), patched it, and relaunched the full population in week 4. Final outcome: 3.2% lift in core engagement, captured 5 weeks earlier than the slow path would have, and we caught the creator regression in 90 minutes instead of the 7+ days it would have taken in the slow rollout.

The thing I take away is that Move Fast is not about skipping safety; it is about engineering safety to be cheap enough that you can move fast and still recover. The rollback gates were the move that made the velocity defensible. I now propose the rollback design before I propose the rollout speed.'

What lands: a real velocity tradeoff named explicitly, the engineering of safety (rollback gates) that made the velocity defensible, an honest naming of the regression that did happen, the recovery time, and a generalised lesson about engineering safety as the prerequisite for velocity. This is the shape of a strong Move Fast answer at Meta.

Framing 2: Be Direct and Respect Your Colleagues

'I want to share a time I disagreed directly with my tech lead on a launch decision. At my previous company I was an E4 engineer and our tech lead had proposed a 6-week multi-stage rollout for a new feed-ranking model. I thought the proposal was too cautious for the risk profile of the launch and that we were leaving four weeks of engagement gain on the table.

The directness move I made was to write a one-page counter-proposal and share it with him before the team meeting, not in the team meeting. The reason: I wanted to disagree on the substance with him directly first, so that any team conversation would be a real discussion rather than a public ambush. The counter-proposal had three sections: the cost of the slow path in foregone engagement, the rollback design that would make a fast path safe, and an explicit list of the things he was right to worry about.

He pushed back hard on two points. The first was the alerting threshold I had picked, which he thought was too lenient; he was right and I tightened it. The second was about creator-segment risk specifically; he was also right and I added a creator-segment alert that I had not had in the original design. The proposal we took to the team was therefore a joint one with both names on it, not a unilateral push from me.

The launch went ahead with the fast path, the creator-segment alert tripped in week 3 exactly as he had predicted, and we recovered in 90 minutes. He brought it up at our 1-1 and said the joint document was the model he wanted to use for future disagreements on the team. The thing I take away is that Be Direct is paired with Respect Your Colleagues for a reason: directness without respect would have been pulling rank in a public meeting, which would have damaged the team. Directness with respect was a private 1-1 with a written counter-proposal, which made the team conversation better and ended with both of us trusting each other more.'

What lands: an explicit naming of the directness move (writing a counter-proposal, sharing privately before public), genuine acceptance of the tech lead's pushback (two specific concessions), a closing observation that distinguishes Be Direct from public bluntness, and a relationship that came out stronger. Meta interviewers grade this kind of story very highly.

Worked Example 2: A Fresh Story for 'Meta, Metamates, Meta' (Company First)

This value is unique to Meta and is graded sharply. The shape requires a real moment where the candidate prioritised company outcome over team outcome, ideally at some cost to their own metrics.

'I want to share a time my team absorbed real cost so that another team could ship something the company needed. At my previous company we were running an internal platform for ML feature engineering, and a sister team building a new recommendation product had a critical dependency on us shipping a new feature-store API by end of quarter. We had not committed to that timeline; our roadmap said next quarter. The company-level priority was the recommendation product; our team's priority was a separate scaling project we had committed to.

The honest read was that both projects mattered for the company, but the recommendation product was on a critical path tied to a public launch and our scaling project was internal. My manager and I talked about it. The default move would have been to push back and say our roadmap was committed, which was technically true. The Meta, Metamates, Meta read was different: the company would be better off if the recommendation product shipped on time, even at cost to our team's quarterly metrics.

We took the work on. I led the API rebuild. We pulled two engineers off the scaling project for six weeks. The recommendation product shipped on time. Our team's scaling project slipped by about a month, and our team's quarterly review noted the miss explicitly; we did not try to hide it. My manager protected the team in the review by surfacing the company-level tradeoff openly: we slipped because we absorbed dependency work for a higher-priority initiative.

The recommendation product went on to be a meaningful contributor to the company that year. Our scaling project landed a month late and the cost was real but recoverable. The thing I take away is that Meta, Metamates, Meta is not about being a martyr; it is about reading the company-level tradeoff honestly when the team-level tradeoff is unfavourable, and being willing to take the team-level cost when the company-level math works out. I now actively look for those tradeoffs at the start of a quarter rather than discovering them in week 8.'

What lands: a real cost paid (the team's project slipped by a month, the quarterly review surfaced the miss), an explicit framing of the tradeoff in company versus team terms, no martyrdom (the manager protected the team transparently), and a generalised behavioural change about reading tradeoffs early. This is the shape of a strong Metamates story.

Red Flags & Green Flags

Green flags (the People round writes 'hire'):

• The candidate names velocity tradeoffs explicitly. Stories where the candidate made an active speed-versus-thoroughness call, with reasoning, score better than stories where the candidate optimised for thoroughness silently.
• Direct-disagreement stories include a private-first beat (1-1 conversation, written counter-proposal) before any public step. Pure-public-disagreement stories are read as performative.
• The candidate accepts pushback genuinely and concedes specific points where the data demanded it. Pure backbone with no concession is read as positional.
• Cross-team stories include real names (anonymised but specific: 'the recommendation team' not 'another team'), which signals the story is real and not constructed for the interview.
• Reflections are behavioural changes, not generic learnings. 'I now propose the rollback design before I propose the rollout speed' is high-signal; 'I learned the importance of communication' is not.

Red flags (the People round writes 'no hire'):

• Bluntness without respect. A candidate who tells a story about 'just being direct' that involves embarrassing a colleague in public, going around someone, or escalating without warning is read as misunderstanding the value badly.
• Velocity claims with no actual velocity tradeoff. 'I moved fast' without naming what was deprioritised or accepted as risk is empty. The interviewer is grading whether the candidate understands that velocity has a cost they paid intentionally.
• 'I would do X' instead of 'I did X'. Hypothetical answers do not count for the People round.
• Defensive posture in response to follow-up questions. Be Direct cuts both ways; the interviewer is implicitly testing whether the candidate can absorb hard questions about their stories without becoming defensive.
• Stories where the candidate consistently chose team or individual benefit over company benefit, especially when probed on Metamates. The order is Meta, Metamates, Meta for a reason.
• Lack of opinion. Meta grades for whether the candidate has formed views about how to work. Stories told as 'I executed against the plan' without showing the candidate's own perspective are read as low-signal.

Mock Interview Walkthrough: A People Round

The following is a simulated 45-minute People round for an E4 software engineer. Interviewer-internal-reaction commentary in italics.

Interviewer: 'Thanks for joining. This is the People round. I am going to ask a few questions, and I am looking for real stories with real specifics. First one: tell me about a time you disagreed with a manager or tech lead and what you did.'

Interviewer mental note: Be Direct probe. I want a real disagreement, a private-first move, genuine acceptance of pushback, and a relationship that survived. Bonus if the candidate names the directness move explicitly.

Candidate: [delivers the feed-ranking story framed for Be Direct, as in Worked Example 1, framing 2.]

Interviewer mental note: very strong. The 'I wrote a one-page counter-proposal and shared it with him before the team meeting, not in it' beat is exactly the kind of directness-with-respect signal we look for. The two specific concessions to his pushback land cleanly. Hire on Be Direct.

Interviewer: 'Tell me about a time you shipped something faster than would have been comfortable, and what tradeoffs you made.'

Interviewer mental note: Move Fast probe. I want a real velocity-versus-X tradeoff named explicitly, with reasoning. Pure 'we shipped fast' without naming the tradeoff is empty.

Candidate: [delivers the same feed-ranking story, but reframes for Move Fast, as in Worked Example 1, framing 1. Acknowledges the overlap with the first answer.]

Interviewer mental note: smart move to flag the overlap. The Move Fast framing is genuinely different from the Be Direct framing of the same story. The 'engineer safety to be cheap enough that you can move fast' insight is unusually mature for E4. The rollback-as-velocity-enabler framing is exactly right. Hire on Move Fast.

Interviewer: 'Walk me through a time you put company priorities above your team's.'

Interviewer mental note: Metamates probe. I want a real cost paid by the team, an honest framing of the tradeoff, and no martyrdom.

Candidate: [delivers the feature-store dependency story from Worked Example 2.]

Interviewer mental note: clean Metamates story. The team paid a real cost (a month slip surfaced openly in the quarterly review), the framing was company-versus-team explicitly, and the closing observation about reading tradeoffs early is generalisable. Hire on Metamates.

Interviewer: 'Tell me about a time you gave hard feedback to a peer. Be specific about how you delivered it.'

Interviewer mental note: another Be Direct probe, now testing peer-level feedback rather than upward disagreement. I want directness paired with respect: said the hard thing, said it in a way the recipient could act on, relationship survived.

Candidate: [delivers a fresh story about giving feedback to a peer who was systematically under-reviewing PRs from junior engineers, including the private 1-1 framing, the specific examples used, the peer's initial defensive reaction, the candidate's follow-up to make sure the feedback landed, and the visible behavioural change three months later.]

Interviewer mental note: very strong. The 'I followed up after the initial defensive reaction' beat is unusual and exactly right; most candidates skip it because it is uncomfortable to recount. The three-month visible change is verifiable. Hire on Be Direct (peer-level), confirmed.

Interviewer: 'Last question. Tell me about something you learned in the past year by working on something you had not worked on before.'

Interviewer mental note: implicit Live in the Future and Build Awesome Things probe. I want curiosity tied to ambition: working on something hard or new, not just on something incremental.

Candidate: [delivers a story about spending a quarter on an embedding-quality project that the candidate had pushed for despite team scepticism, the technical learning involved, and the eventual integration into the production ranking pipeline.]

Interviewer mental note: solid. The team-scepticism beat is a Live in the Future signal (working on what is hard, not what is comfortable). The integration into production is the Build Awesome Things signal. Hire on Live in the Future and Build Awesome Things.

Debrief outcome: Strong hire across all five values probed. The People round writes 'strong hire'. Likely offer.

How to Prepare in 8 Hours

• Hour 1: Read Meta's published values and recent Mark Zuckerberg communications about company direction (the 2022 reorganisation, the Year of Efficiency framing, the AI focus). Internalise the velocity-and-direct-feedback culture.
• Hour 2: For your existing story bank, identify which stories can demonstrate Move Fast and Be Direct. These are the two highest-frequency probes. You need at least two strong stories for each.
• Hours 3 to 5: Write tailored framings for your top 6 stories (one per 30 minutes). For Move Fast stories, name the velocity tradeoff explicitly. For Be Direct stories, include the private-first beat and the specific concession to pushback.
• Hour 6: Build one strong Metamates story. This value is uniquely Meta's and is graded sharply. The shape needs a real cost paid by your team for the company's benefit, framed honestly.
• Hour 7: Practice receiving hard follow-up questions non-defensively. Be Direct cuts both ways and the People round will probe. Practice saying 'that is a fair pushback, let me think about it' rather than reflexively defending.
• Hour 8: Mock interview, with the partner asked to push back hard on at least one of your stories. Practice the receive-feedback posture.

Bridge to the Next Lesson

This lesson covered Meta, where velocity and direct disagreement are explicit cultural anchors. The next lesson, Apple: Craftsmanship and Collaboration, covers a culture in many ways the opposite: a company where attention to detail and craftsmanship are explicit, where iteration cycles can be longer, and where the 'this is not ready yet' instinct is a positive signal rather than a velocity drag. The contrast is instructive. Some of your stories will reframe well across both companies; others will only fit one.