Talent & Culture Strategy at Meta: Cull Low Performers, Buy Brilliance

This article is part of MokaHR's Talent & Culture Strategy series, which profiles how leading companies build their people strategies.

In January 2025, Mark Zuckerberg posted a memo on Meta's internal forum. "I've decided to raise the bar on performance management and move out low performers faster," he wrote. Within weeks, roughly 3,600 employees — about 5% of the workforce — had been notified of termination, all on performance grounds. Six months later, Zuckerberg was engineering the most expensive individual talent acquisition in technology history: a compensation package reportedly worth $1.5 billion over six years to secure a single AI engineer from a competitor startup.

These two events are not contradictions. They are the same strategy, expressed from opposite ends. Meta's talent and culture strategy in 2025 and 2026 is defined by a single operating principle: obsessive concentration of quality. Remove the bottom. Pay whatever it takes for the top. Everything else — the Ohana-style culture investments, the DEI programmes, the middle layers — is secondary to talent density.

Meta's employer brand recognition is reinforced by consistent appearances on Fortune's 100 Best Companies to Work For list and Glassdoor's Best Places to Work rankings, though the company's aggressive performance culture has attracted increasing scrutiny. In 2025, Meta was named one of LinkedIn's Top Companies in the United States for career growth.

How does Meta attract and hire talent?

Two hiring tracks, two different companies

Meta runs what is effectively two distinct talent acquisition operations simultaneously, and understanding the difference between them is essential to understanding the company's overall HR strategy.

The first track — covering the vast majority of the company's ~79,000 employees — follows a rigorous but recognisable hiring process. Candidates move through resume screening, a recruiter phone screen, a technical or case interview, and a final loop of four to six structured interviews covering technical skills, behavioural competencies, and cultural alignment. The company uses standardised scoring rubrics and calibration sessions to minimise interviewer bias and maintain consistency at scale.

The second track, created in 2025 for Meta Superintelligence Labs, operates by entirely different rules. There are no job postings. Zuckerberg maintains what sources describe as a personal list of AI researchers he intends to recruit. Outreach is direct, often coming from Zuckerberg himself. Compensation packages are individually negotiated and, in some cases, reach hundreds of millions of dollars. Andrew Tulloch, co-founder of Mira Murati's AI startup Thinking Machines Lab, reportedly received a package worth $1.5 billion over six years — which, if accurate, represents the single most expensive individual hire in the technology industry's history. OpenAI chief scientist Shengjia Zhao, a co-creator of ChatGPT, was recruited to serve as Meta Superintelligence Labs' chief scientist. Sam Altman, OpenAI's CEO, publicly acknowledged that Meta had offered his staff signing bonuses of $100 million and above.

The recruitment pitch for the Superintelligence Labs is not primarily financial. Meta offers elite researchers small, autonomous teams, minimal bureaucracy, and direct reporting lines to Zuckerberg — a deliberate counter-positioning against perceived organisational bloat at Google DeepMind and even OpenAI. The message is explicit: you will work on hard problems, with the best people, with the resources of a company generating $200 billion in annual revenue, without the friction of a large organisation.

This dual-track model creates internal tension. In October 2025, approximately 600 roles were cut from Meta's FAIR AI research division and AI infrastructure teams — many of them senior engineers who were not part of the elite Superintelligence Labs cohort. The disparity in compensation, autonomy, and resource access between the two groups is significant and, according to multiple industry analysts, carries real cultural risk.

The performance bar as a recruiting tool

Meta's approach to performance management has become inseparable from its talent acquisition strategy. In January 2025, Zuckerberg introduced a revised performance review system with five rating tiers: Outstanding (top 20% of employees), Excellent (70%), Meets Expectations, Needs Improvement (~7%), and Not Meeting Expectations (~3%). Outstanding performers are eligible for bonuses worth up to 300% of pay. Those rated "Not Meeting Expectations" receive no bonus and face accelerated performance management processes.

The logic is explicit: by exiting low performers faster and more aggressively, Meta creates headcount budget and cultural signal to hire better replacements. As Zuckerberg told managers in a note, the performance-based cuts are "aimed at ensuring the company has the strongest talent and is able to bring new people in." This is performance management as talent refresh, not cost reduction.

The April 2026 announcement of a further 8,000 job cuts — roughly 10% of the workforce, effective May 2026 — follows the same logic on a larger scale. Chief People Officer Janelle Gale's memo to staff described the reductions as necessary to "drive efficiencies and help offset the company's other investments," pointing directly to Meta's commitment to spend $115–135 billion in capital expenditure in 2026, primarily on AI infrastructure.

[Interlink Slot A — contrasting approach:] This concentration-over-breadth philosophy contrasts sharply with how Shopify has built its talent strategy around radical subtraction — removing management layers, eliminating bureaucracy, and trusting small autonomous teams — a different expression of the same underlying conviction that talent density matters more than headcount.

Retention under pressure

Retaining top AI talent is proving more complex than recruiting it. Meta has invested in retention through compensation, autonomy, and what it frames as a mission — becoming the first company to build artificial superintelligence. But early returns are mixed. Yann LeCun, Meta's chief AI scientist for 12 years and one of the most influential figures in deep learning, departed in November 2025 after reportedly being asked to report to the newly installed Alexandr Wang. In his departure interview with the Financial Times, LeCun called the restructuring badly managed and warned that "a lot of people have left, a lot of people who haven't yet left will leave." He subsequently raised $1 billion to found AMI Labs in Paris.

OpenAI's retention rate dipped to 67% in 2025 according to SignalFire's 2025 AI Retention Report — partly attributed to Meta's poaching. Meta's own retention data for its Superintelligence Labs cohort is not yet public, but the structural question is real: can extraordinarily high cash compensation substitute for the mission-driven culture that retains researchers at Anthropic (80% retention) and Google DeepMind (78%)?

📄 2025 AI Recruitment Casebook Meta's Superintelligence Labs represents the extreme end of AI talent competition — but companies across every industry are now competing for AI-literate professionals using tools that didn't exist three years ago. See how companies in tech, pharma, financial services, and seven other industries are adapting their recruitment approaches in MokaHR's AI Recruitment Report. Download the free report →

Trusted by: Tesla · NVIDIA · McDonald's · Nestlé · Schneider Electric

How does Meta develop and manage employee performance?

Learning and development in an output-first culture

Meta's approach to employee development has always been high-velocity rather than structured. The company does not run formal leadership academies or multi-year rotation programmes. Development at Meta happens through increasing ownership — engineers and managers are given scope and accountability quickly, and performance is measured by output rather than tenure or process.

The company does invest in formal learning infrastructure. Meta offers access to internal learning platforms covering technical skills, management development, and functional competencies. All new managers complete a structured onboarding programme, and the company runs targeted leadership programmes for managers moving into senior individual contributor or people management roles.

The performance review system drives development as much as it evaluates it. The five-tier rating model — updated in early 2026 — provides explicit guidance on what distinguishes an "Excellent" performer from an "Outstanding" one. The distinction matters: the bonus differential between the two tiers is significant, and only "Outstanding" performers are considered for the most senior technical and leadership roles. This creates a clear, measurable signal for what Meta values: impact on the company's most important problems, not effort, process adherence, or seniority.

From annual to continuous feedback

In January 2026, Meta updated its performance programme to place greater emphasis on frequent feedback and recognition rather than a single annual review. According to a Meta spokesperson, the change was designed to "simplify and reward outstanding performance" while providing more regular signals to employees about where they stand. Managers are expected to deliver ongoing performance signals throughout the year, not just at formal review points.

The revised system also introduced more explicit pay multipliers. Employees rated "Excellent" — the expected outcome for around 70% of the workforce — are eligible for a bonus of 115% of target. "Outstanding" performers can receive bonuses of up to 300%. The "Needs Improvement" tier results in bonuses capped at 50%, and "Not Meeting Expectations" renders employees ineligible for bonuses entirely and places them on accelerated performance improvement or exit processes.

[Interlink Slot B:] Meta's shift toward output-based, continuous feedback has parallels with how Adobe dismantled annual performance reviews in favour of regular check-ins through its Check-in programme — a transformation that contributed to a 30% reduction in voluntary turnover.

The takeaway from Meta's development model: it invests in people who are already performing at a high level, not in those who need significant uplift. The company's philosophy is that development is an accelerant for top talent, not a rescue mechanism for underperformers.

What can HR leaders learn from Meta's talent strategy?

Meta's approach is not fully replicable. The compensation packages in its Superintelligence Labs — reportedly reaching hundreds of millions of dollars for individual hires — are available to a vanishingly small number of companies. The strategic context — competing to build artificial superintelligence against a handful of well-resourced rivals — is unique.

But the underlying principles translate more broadly.

Define what a top performer looks like, then price it explicitly. Meta's five-tier system is notable not just for its performance differentiation but for its transparency. Employees know what "Outstanding" means and what it pays. Most companies' performance reviews produce grades without clear market anchors — which means top performers often leave not because they are dissatisfied with their rating, but because they cannot see the financial consequence of it. AI-powered compensation analytics can help HR teams model what their top performers are worth on the open market, and close that gap before competitors identify it.

Use performance management as active talent refresh, not just evaluation. Meta's explicit goal in its performance cycle is not just to score employees, but to create openings for stronger hires. This requires an ATS that can manage the downstream hiring pipeline that follows a performance-based exit cycle — tracking open roles created, time-to-fill, and quality of replacement hires. Structured hiring workflows with integrated pipeline analytics make this kind of systematic talent refresh manageable at scale.

Protect elite talent from organisational overhead. Meta's recruitment pitch for Superintelligence Labs — small teams, direct access to leadership, minimal bureaucracy — is an HR design choice, not just a compensation one. Companies at any scale can apply this principle: identify your highest-impact roles and design away the overhead that slows those people down.

Be honest about the cultural cost of a two-speed organisation. Meta's internal divide between the Superintelligence Labs cohort and the broader AI organisation is a real risk. The exit of Yann LeCun and the October 2025 cuts to FAIR are early warning signals. Any HR director considering a similar tiered model should plan explicitly for how to manage morale, equity perception, and retention across both tiers.

[Interlink Slot C:] For a contrasting model — one that builds high performance through culture and development rather than aggressive performance cuts — Google's Project Oxygen framework and internal mobility infrastructure offers a detailed counterpoint worth studying.

What is it like to work at Meta?

The employee experience at Meta has changed substantially over the past three years. The company that declared a "Year of Efficiency" in 2023, cut 19,000 roles, and followed with further performance-based reductions in 2025 and structural layoffs in 2026 is not the same employer it was during its pandemic-era expansion.

The physical work environment is a genuine differentiator. Meta's Menlo Park headquarters — designed by Frank Gehry — includes extensive amenity infrastructure: fitness facilities, healthcare services, free meals, dry cleaning, and a range of recreational spaces. The campus is designed to encourage serendipitous cross-functional interaction, consistent with Meta's belief that much of the best work happens in informal conversation between engineers and designers.

Flexible working remains a priority for many employees, though Meta has progressively tightened its return-to-office expectations. Most corporate roles require in-person attendance for a portion of the week, and the company tracks badge data to monitor compliance.

Where Meta has become more complicated as an employer is in the psychological experience of working in a performance culture defined by visible peer comparison. The explicit performance distribution — with a mandated percentage of employees in each tier — means employees are not just evaluated on their own output but implicitly ranked against colleagues. The mid-2025 review cycle, which asked managers to identify 15–20% of employees as "below expectations" (up from 12–15% in the prior cycle), drew significant employee concern, with critics noting that such forced distributions can produce outcomes that reflect organisational politics as much as individual performance.

Meta ended its major DEI programmes in January 2025, disbanding its DEI team, closing its "Diverse Slate Approach" for hiring, and terminating its diversity supplier programme. Chief People Officer Janelle Gale, who announced the changes, cited the evolving legal and policy landscape as the primary driver. The decision sparked significant internal and external debate.

The honest assessment: Meta offers exceptional compensation, significant technical challenge, and access to infrastructure resources few companies can match. What it does not offer is the sense of psychological safety that comes from knowing your position is stable. For the right type of high-output professional, that is an acceptable trade.

Frequently asked questions

How does Meta hire employees? Meta uses structured interviews, rigorous technical assessments, and values-based screening for most roles. Candidates move through a phone screen, a technical or case interview, and a final loop of four to six interviewers using standardised scoring rubrics. For its AI Superintelligence Labs, Meta operates a separate direct-recruitment track for elite researchers — with direct outreach from senior leadership and compensation packages negotiated individually, in some cases reaching hundreds of millions of dollars over multi-year agreements.

What is Meta's performance management approach? Meta uses a five-tier output-based review system: Outstanding (top ~20%), Excellent (~70%), Meets Expectations, Needs Improvement (~7%), and Not Meeting Expectations (~3%). Outstanding performers are eligible for bonuses worth up to 300% of target pay. In early 2025, Zuckerberg introduced accelerated performance-based cuts, removing roughly 3,600 employees rated in the lowest tiers. A further 8,000 role reductions were announced in April 2026 as part of a broader AI-focused reallocation of resources.

How many people work at Meta? Meta had approximately 78,865 employees at the end of fiscal year 2025. Following the April 2026 announcement of roughly 8,000 planned cuts (effective May 2026), headcount is expected to fall to approximately 70,000 — representing a return to near the post-2023-efficiency-cuts level, but with a workforce now far more concentrated in AI and engineering.

What is Meta Superintelligence Labs? Meta Superintelligence Labs is an elite AI research division established in mid-2025. It is led by Alexandr Wang (formerly of Scale AI, brought in via a $14.3 billion investment) and chief scientist Shengjia Zhao (co-creator of ChatGPT). The division operates with small teams, direct access to Zuckerberg, and a mission to build artificial superintelligence. It is structurally separate from Meta's broader AI organisation and is designed to attract the world's most elite AI researchers through premium compensation and startup-like operating conditions.

See also

• Talent & Culture Strategy at Google: Data-Driven People Ops at Scale

• Talent & Culture Strategy at OpenAI: Mission, Equity, and the AI Talent Race

• Talent & Culture Strategy at Anthropic: Safety-First Hiring in a Frontier AI Lab

The gap between Meta's Superintelligence Labs' compensation and what most organisations can offer will never close. But the structural principles — calibrated performance tiers, talent refresh through managed exits, elite team insulation from bureaucracy — are available to any HR leader willing to apply them. See how MokaHR helps organisations implement rigorous, data-driven hiring pipelines built for exactly this kind of performance-first culture. Book a personalised demo →