How Meta's CTO thinks about leadership, communication, and building at scale

Executive overview

Most people credit great products to great ideas. Boz argues the real bottleneck is communication: if the idea didn't land, that's on you, not the audience. At Meta, this principle shapes everything from how teams get updates to how Mark Zuckerberg processes feedback.

The core insight: communication isn't a soft skill layered on top of the job — it is the job.

Asking your manager for help is underrated

• Most people resist asking because they want to prove they can do it alone — but the job is to get it done, not to do it yourself
• Light-touch updates ("here's where I am, no response needed") keep your manager in the loop without demanding their time
• When you are blocked, draft the specific email or question you need sent — make it cheap for your manager to help
• Nobody wants you to succeed more than your manager, because your success makes their life easier
• Ask new managers early: "How do you like to receive updates?" — every manager is different
• Meta used "HPMs" (highlights, people, me) — a structured weekly note sent up the chain

Radical transparency has a cost

• Open information maximises what talented people can do; blocked information wastes your most valuable resource
• The real bottleneck is incoming signal, not access — senior hires need a system for triaging noise
• Meta is top-down on big ideas (Mark, Cox, Javi set the agenda) but gives wide latitude on execution within those constraints
• Clear guardrails matter: people need to know where they have freedom and where they do not
• Leaks are a violation of team trust — equivalent to calling out a play to the opposing team

When to go deep as a leader

• There is no universal rule — some weeds are the hinge of success or failure; others just need to be done
• Mark Zuckerberg's "eye of Sauron": when something is the most important thing, no detail is too small
• Being inside the eye of Sauron too long burns people out; being too far from it (the Shire) kills visibility
• The two best career positions: carrying critical work nobody is watching, or being on the most important thing
• If you are not learning and not engaged, move on — compound interest on skills takes years to show

Communication is the job

• Impact on the world happens exclusively through artifacts and verbalizations that affect other people
• "I had that idea" means nothing; if it didn't break through, that is on the communicator, not the audience
• Org charts, silence, facial expressions, and clothing are all communication — you cannot not communicate
• Multimodality matters: say it in an all-hands, then write it up — different people absorb information differently
• Address people's fears first; they will not hear your conclusions if they do not accept your premise
• Extreme ownership: when something goes wrong, ask what you could have communicated differently

Meta's turnaround: what actually happened

• Largest single-day stock drop in history, followed 18 months later by the largest single-day gain
• Lesson from Lou Holtz: you are never as good as they say when winning, never as bad when losing
• The core failure was not communicating the investment thesis (AI + Reality Labs) to the market clearly enough
• COVID pulled forward growth that then reverted to its original trajectory — the economics stopped working
• Much of the workforce had never seen a downturn; the correction was also generational, not just cyclical
• AI research (FAIR) is the second most-cited lab in AI behind Google — the work was always there

On building the News Feed and product conviction

• The first algorithmic ranked feed of any social network — and the first consumer AI on a website
• Users said they hated it while immediately doubling their usage: revealed preference over stated preference
• "You're choosing your customers as much as they're choosing you" — early adopters can hold you hostage
• Distinguish between the thing being wrong and the details being wrong; that diagnostic is an art
• Hand tracking on Quest headsets was called impossible internally — Boz forced it and held high standards

Career advice

• Early career: optimise for learning, not title — variety compounds like interest and pays off later
• Staying in one place gets you promoted faster early, but can leave you domain-specialist trapped mid-career
• Give new areas six months; if you cannot find the passion, move on — but you can have a broader palette than you think
• Boz moved every six months early on (News Feed, anti-spam, infrastructure, bootcamp, messaging, ads) and went vertical once he found a deep role

Identity threat and curiosity as a leadership tool

• Worst behaviour always emerges under identity threat — when who you are feels in question, you defend it with everything
• The fix is not suppressing the reaction but replacing it with genuine curiosity: "Fascinating — tell me more about why you see it that way"
• This disarms conflict, opens the other person, and often changes your own mind
• People remember how you made them feel, not what you said — embarrassing moments from decades ago still teach