In 1968, Robert Rosenthal and Lenore Jacobson ran an experiment at a San Francisco elementary school that would reshape how psychologists thought about expectation and performance. They told teachers that certain students had been identified by a special test as "intellectual bloomers" — kids about to experience a dramatic surge in academic ability. The test was fake. The students were randomly selected. But by the end of the year, the "bloomers" showed significantly higher IQ gains than their peers. Teachers had unconsciously given these students more attention, more challenging material, warmer feedback, and more patience. The expectation created the reality. Rosenthal and Jacobson called it the Pygmalion Effect.
The Golem Effect is its dark twin.
Named after the golem of Jewish folklore — a creature brought to life by a rabbi but constrained by its maker's limited expectations of what it could become — the Golem Effect describes what happens when low expectations create low performance. Not through incompetence. Not through lack of talent. Through the systematic withdrawal of the conditions that performance requires. When a manager expects someone to fail, they give less feedback, fewer growth opportunities, less autonomy, less access to resources, less benefit of the doubt. The employee, deprived of the raw materials of development, underperforms. The manager sees the underperformance and thinks: "I knew it." The prophecy fulfils itself not because it was accurate but because it was acted upon.
The mechanism is brutal in its precision. Rosenthal's subsequent research documented the four-factor framework through which expectations transmit: climate (warmth, attention, eye contact), input (more challenging material for high-expectation subjects), output (more opportunities to respond and contribute), and feedback (more specific, more constructive, more frequent). When expectations are high, all four factors increase. When expectations are low, all four decrease — often unconsciously. The manager who has labelled an employee "not leadership material" stops inviting them to strategic meetings. The manager who has decided a team is "B-level" allocates fewer resources and less executive attention. The manager who has concluded a new hire "probably won't work out" gives less onboarding support and checks in less frequently. Each withdrawal is individually small. Collectively, they construct an environment where underperformance is the rational adaptive response.
The Israeli Defense Forces provided one of the most rigorous demonstrations. In a 1982 study, Dov Eden randomly assigned leadership-expectation labels to incoming trainees in a military course. Instructors were told that certain trainees had been identified as having "high command potential" based on psychological assessments. The assessments were fabricated. The "high potential" trainees outperformed their peers significantly — not because of innate ability but because the instructors taught them differently. They received sharper corrections, more individual coaching, and more demanding exercises. The trainees labelled average received average instruction and produced average results. The expectation did not predict performance. It manufactured it.
In corporate environments, the Golem Effect operates at every level. A CEO tells the board that the European division is "non-strategic." Resource allocation shifts toward the "strategic" divisions. The best talent transfers out of Europe because career advancement follows resources. The European leadership team, stripped of investment and top performers, delivers declining results. The CEO reviews the numbers and says, "This is why we deprioritised Europe." The circularity is invisible to everyone inside it. The label created the conditions that validated the label.
Startup culture is especially fertile ground. A founder who tells the team "we're probably not going to make it" is not being honest. They are triggering the Golem Effect at organisational scale. Top performers start interviewing elsewhere. Middle performers reduce effort because the expected outcome no longer justifies personal sacrifice. The founder's prediction becomes a coordination signal: disengage. The startup's decline accelerates — not because the market was unwinnable but because the founder's expectation withdrew the commitment that winning required.
The most insidious feature: the Golem Effect operates below conscious awareness. Managers do not decide to withdraw resources from low-expectation employees. They simply allocate more to high-expectation ones. Teachers do not choose to call on certain students less. They naturally gravitate toward the students who "seem" most engaged — the ones they have already labelled as high-potential. The asymmetry feels like merit-based allocation. It is expectation-based allocation wearing the mask of objectivity. The Golem Effect does not require malice. It requires only assumptions — and humans carry assumptions into every interaction. The antidote: assume competence until proven otherwise.
Section 2
How to See It
The Golem Effect is visible in the gap between an individual's or team's actual capability and the performance that the surrounding environment allows them to demonstrate. The tell is not low performance itself — it is low performance accompanied by systematically reduced inputs: fewer opportunities, less coaching, less investment, less attention.
You're seeing the Golem Effect when underperformance is accompanied by underinvestment — and the underinvestment preceded the underperformance.
Management & Leadership
You're seeing the Golem Effect when a manager describes a direct report as "not ready for the next level" but cannot point to any structured development they have provided. The employee has not been given stretch assignments, has not been invited to cross-functional meetings, has not received the coaching that would develop the skills the manager says are missing. The manager is not observing a ceiling. They are building one — and then standing back and describing it as the employee's limitation.
Organisational Design
You're seeing the Golem Effect when a business unit consistently labelled as "cost centre" or "non-core" underperforms relative to units labelled "strategic." The labelling triggers a cascade: less executive attention, smaller budgets, weaker talent pipelines, lower retention of existing talent. The unit's declining results then justify the label. The diagnostic: compare resource allocation per employee across units. If the "non-core" unit receives systematically less investment per head, the Golem Effect — not inherent capability — is the most parsimonious explanation for the performance gap.
Venture Capital & Startups
You're seeing the Golem Effect when a VC partner writes off a portfolio company after a single bad quarter and reduces engagement — fewer board calls, less intro flow, no follow-on discussion. The company, deprived of the support network that other portfolio companies receive, struggles to recruit executives, close partnerships, and raise follow-on capital. The partner reviews the company twelve months later and says, "I knew that one wasn't going to work." The withdrawal manufactured the outcome the partner is now claiming to have predicted.
Education & Coaching
You're seeing the Golem Effect when a coach or teacher gives up on a student early and redistributes attention to "higher-potential" individuals. The student receives less correction, less challenge, less encouragement. Their performance stagnates or declines. The coach points to the stagnation as evidence of limited ability. Rosenthal's research showed this pattern operates even when the initial ability assessment was random — the expectation, not the ability, determined the trajectory.
Section 3
How to Use It
The Golem Effect's operational lesson is defensive: your expectations are not observations — they are interventions. Every time you label someone as low-potential, you are not describing reality. You are beginning to construct it. The defence is building systems that hold resource allocation and opportunity access constant regardless of the manager's private assessment.
Decision filter
"Before labelling a person, team, or initiative as low-potential: Have I provided the same resources, coaching, opportunities, and attention that my high-expectation bets received? If not, I am measuring the effect of my expectation, not the capability of the subject."
As a founder
The Golem Effect is a founder-level risk because the founder's expectations propagate through the entire organisation. When you tell a team their project is "low priority," you are not just communicating resource constraints. You are signalling that the humans on that team are lower priority — and they will calibrate their effort accordingly. The fix is not motivational language. It is resource discipline. If a project is genuinely lower priority, communicate the strategic reasoning while maintaining clear development paths and feedback loops for the team. If you cannot invest in a team, disband it and redeploy the people. The worst outcome is a team that exists but operates under the Golem Effect — low expectations producing low results that justify the low expectations.
The hiring version is equally dangerous. A founder who hires someone they privately consider "good enough for now" will unconsciously withhold the onboarding, mentorship, and opportunity access that would allow that hire to exceed the initial assessment. If someone is worth hiring, they are worth investing in as if they might be exceptional. If they are not worth that investment, they are not worth hiring.
As a team lead
Audit your own behaviour for Golem markers. Track — literally write down — how you allocate your one-on-one time across your team. Count how many stretch assignments go to each person. Note who gets invited to leadership meetings and who doesn't. Note who receives coaching conversations and who receives only status updates. If the allocation pattern mirrors your private ranking of team members by potential, the Golem Effect is operating. Your top-ranked people are getting the conditions that produce top performance. Your bottom-ranked people are getting the conditions that produce bottom performance. You are not measuring ability. You are measuring your own investment.
The structural fix: standardise inputs before measuring outputs. Every team member gets the same onboarding, the same access to challenging work, the same frequency of coaching. After six months of equal investment, performance differences are more likely to reflect actual capability rather than differential expectation. This is harder than it sounds because the brain resists investing equally — it wants to invest where the return seems highest. But the "return" you're using to make that judgment is itself shaped by your prior investment. You are optimising a loop, not evaluating a truth.
As a decision-maker
Watch for the Golem Effect in strategic planning. Business units, product lines, and geographic regions that receive the label "declining," "legacy," or "non-strategic" will decline — partly because the label redirects resources away from them. The question is not whether the label is currently accurate. The question is whether the label is causing the trajectory it describes. Intel in 2005 was correct that the PC market was mature. But labelling the PC division as non-strategic accelerated the talent drain and underinvestment that made the division's decline steeper than the market demanded. The strategic frame became a Golem frame — the expectation didn't just describe the future, it constructed it.
Before labelling any unit as non-strategic, calculate what the unit would need to compete effectively. If you are unwilling to provide those resources, divest or wind down. But do not maintain the unit while starving it — that is the Golem Effect operating at the organisational level, producing the failure that justifies the divestment you should have made earlier.
Common misapplication: Using the Golem Effect to argue that all low-performance assessments are expectation artefacts. They are not. Some employees genuinely lack the skills for their role. Some teams face market headwinds that no amount of investment would overcome. The Golem Effect is the specific pattern where underperformance is accompanied by reduced investment — where the expectation preceded and shaped the outcome rather than following from it. The diagnostic is temporal: did the low investment come before or after the low performance? If before, the Golem Effect is the leading hypothesis. If after, the performance problem may be real and the reduced investment may be rational triage.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
The leaders who neutralise the Golem Effect share a structural discipline: they build systems that invest in people based on standardised processes rather than private expectations. They recognise that their own assessments are not neutral observations — they are interventions that shape the outcomes they purport to measure.
Hastings built Netflix's talent philosophy around a principle that is, at its core, a structural defence against the Golem Effect: high talent density eliminates the conditions where differential expectation operates. The "keeper test" — "Would I fight to keep this person?" — cuts both ways. If you wouldn't fight to keep someone, you're creating a Golem environment: that person sits under low expectations, receives diminished investment, and underperforms. The test forces the binary. You are either operating at the level where we would fight to keep you, or you receive a generous severance and leave. There is no middle zone where someone sits under low expectations receiving diminished investment. The cruelty of this system is debatable. Its effectiveness against the Golem Effect is not. When every person in the room knows they are there because leadership expects exceptional work from them, the four-factor model runs in the positive direction. The Golem Effect cannot operate where low-expectation retention does not exist.
Bezos combats the Golem Effect at the point of entry: the bar raiser. Amazon's hiring process requires that every offer include a "bar raiser" — an interviewer from outside the hiring team whose job is to ensure the candidate raises the average capability of the organisation. The bar raiser can veto a hire. The mechanism is explicit: assume competence until proven otherwise, but set the bar high enough that you never hire someone you would unconsciously write off. The Golem Effect thrives when managers hire "good enough for now" and then withhold the investment that would let that hire exceed the assessment. The bar raiser prevents that by ensuring every hire meets a standard where the default expectation is high. If you wouldn't invest in someone's development as if they might be exceptional, don't hire them. The antidote to Golem is structural: raise the bar at hiring so that low-expectation retention never enters the system.
Section 6
Visual Explanation
The top section maps the self-fulfilling loop that defines the Golem Effect. A manager forms a low expectation — perhaps from a first impression, a résumé detail, a colleague's offhand comment, or a single weak interaction. That expectation transmits through four channels: climate (less warmth and attention), input (easier, less challenging work), output (fewer opportunities to contribute and be visible), and feedback (less specific, less frequent, more evaluative). The employee, operating in this impoverished environment, underperforms — not because of limited capability but because the conditions for strong performance have been withdrawn. The manager observes the underperformance and concludes their initial assessment was correct: "I knew it." The loop closes. The next cycle begins with the expectation reinforced.
The feedback arrow on the right — the "I knew it" confirmation — is the mechanism that makes the Golem Effect self-sustaining. Each cycle produces real underperformance that the manager interprets as validation rather than consequence. Breaking the loop requires intervening at the investment stage: standardising the resources, opportunities, and attention that every individual receives before measuring their output.
The bottom panel shows the Golem Effect alongside its mirror image, the Pygmalion Effect. The mechanism is identical — expectations shape investment, investment shapes performance, performance confirms expectations. The only difference is the direction. High expectations drive the cycle upward. Low expectations drive it down. The implication is uncomfortable: every time a manager identifies a "star performer," they may be partially observing the Pygmalion Effect. Every time they identify a "low performer," they may be partially observing the Golem Effect. The performance they are measuring is never fully independent of the expectations they brought to it.
Section 7
Connected Models
The Golem Effect sits at the intersection of leadership psychology, organisational behaviour, and self-regulation theory. The connected models explain how low expectations form, how they transmit, how they compound, and what structural conditions either amplify or neutralise them.
Tension
Pygmalion Effect
The Pygmalion Effect is the Golem Effect's mirror. Both demonstrate that expectations shape performance through the same four-channel mechanism — climate, input, output, feedback. The tension is in the asymmetry: the Pygmalion Effect is celebrated, but the Golem Effect is rarely acknowledged. Organisations invest heavily in identifying "high-potential" employees and creating programmes to accelerate their development. Few organisations audit whether the employees not labelled high-potential are receiving systematically less investment. The Pygmalion-Golem pair reveals that talent identification is partly talent creation — and that every system designed to accelerate top performers may simultaneously be decelerating everyone else.
Reinforces
Self-fulfilling Prophecy
The Golem Effect is a specific instance of the broader self-fulfilling prophecy, first described by Robert K. Merton in 1948. Merton's framework: a false belief generates behaviour that makes the belief come true. The Golem Effect specifies the mechanism — the four-channel withdrawal — and the domain — expectation-performance relationships in hierarchical settings. The reinforcement is bidirectional: every Golem Effect outcome strengthens the belief in the prophecy, which strengthens the expectation, which strengthens the withdrawal. The loop accelerates until the performance gap becomes so large that reversal is practically impossible without external intervention.
Reinforces
[Radical Candor](/mental-models/radical-candor)
Radical candor — Kim Scott's framework of caring personally while challenging directly — is a structural counter to the Golem Effect. The Golem loop thrives when low-expectation individuals receive less feedback, and what they receive is generic and evaluative rather than developmental. Radical candor demands the opposite: give everyone direct, specific, actionable feedback. The low-expectation individual who receives "you're not ready for the next level" without developmental input is in a Golem trap. The same individual who receives "here's the specific skill gap, here's how to close it, here's the stretch assignment to practice" has a path out. Radical candor does not eliminate the Golem Effect — it breaks the feedback channel that perpetuates it by ensuring that even people you have doubts about receive the information they need to improve.
Section 8
One Key Quote
"When we expect certain behaviors of others, we are likely to act in ways that make the expected behavior more likely to occur."
— Robert Rosenthal and Lenore Jacobson, Pygmalion in the Classroom (1968)
The quote is deceptively simple. It contains the entire Golem Effect in a single sentence: expectations are not passive observations. They are active forces that shape the behaviour they purport to predict. The teacher who expects a student to struggle gives that student less time, less challenge, less encouragement — and the student struggles. The manager who expects a hire to underperform gives that hire less onboarding, less coaching, less opportunity — and the hire underperforms. The VC who expects a portfolio company to fail reduces engagement, intro flow, and follow-on support — and the company fails.
The mechanism is not conscious intent. Rosenthal and Jacobson's most important finding was that the teachers genuinely believed they were treating all students equally. They were not. The expectation operated below the threshold of awareness, shaping micro-behaviours — a warmer tone here, a harder question there, an extra thirty seconds of patience — that accumulated into dramatically different outcomes. The Golem Effect does not require a villain. It requires only a human being carrying an assumption into a relationship where they hold power over the other person's environment. The assumption does the rest.
The operational implication is that awareness alone is insufficient. Knowing about the Golem Effect does not prevent it — just as knowing about confirmation bias does not prevent you from seeking confirming evidence. The fix is structural: standardise inputs, audit allocation, create feedback mechanisms that surface differential treatment, and build cultures where the low-expectation individual has the psychological safety to name the pattern. The quote describes the problem. The solution requires architecture.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
The Golem Effect is the single most underdiagnosed performance problem in organisations. Every company has employees who are privately written off but not formally managed out. They sit in roles receiving diminished investment — less coaching, fewer opportunities, less attention from leadership — while the organisation waits for them to leave or provides just enough performance documentation to justify termination. The intervening months or years are pure Golem territory: the employee underperforms because the conditions for performance have been withdrawn, and the underperformance is recorded as evidence of the employee's inadequacy rather than the system's failure.
The most dangerous version operates at the team and business-unit level. When a CEO labels a division "non-strategic," the Golem cascade begins immediately. Top talent transfers out because career progression follows strategic priority. Budget allocation shifts. Executive attention redirects. The division's leadership, stripped of resources and support, delivers results that decline in proportion to the withdrawal. Twelve months later, the CEO shows the board a slide titled "Performance by Business Unit" and the non-strategic division is at the bottom — where the CEO's expectation placed it. The slide reads as evidence. It is actually a mirror.
Startup founders trigger the Golem Effect through language that they mistake for transparency. "I'm not sure this is going to work." "We need to be realistic about our chances." "The market might not be there." Each statement is intended as honest assessment. Each operates as an expectation signal that withdraws commitment. The team hears the founder's doubt and calibrates accordingly — reducing risk-taking, exploring external opportunities, lowering personal investment. The founder's "honesty" manufactures the outcome the founder feared. The alternative is not delusional optimism. It is the discipline to communicate uncertainty without communicating defeat. "The path is hard and the outcome is uncertain, but we are the team that can find it" is a fundamentally different expectation signal than "we're probably not going to make it."
The Golem Effect compounds in underrepresented populations. When bias — conscious or unconscious — creates lower baseline expectations for certain demographics, the four-channel withdrawal follows: less challenging assignments, less coaching, less airtime, less developmental feedback. The performance gap that emerges is then attributed to the demographic rather than to the expectation-driven investment gap that produced it. This is not a diversity argument. It is a Golem Effect argument. The mechanism is identical regardless of the demographic variable. But the baseline expectations that initiate the loop are not randomly distributed — they cluster around the biases the manager carries, which means the Golem Effect disproportionately impacts the people who are already navigating structural disadvantage.
Section 10
Test Yourself
The scenarios below test whether you can identify the Golem Effect operating in professional contexts — and distinguish it from legitimate performance assessment, rational resource allocation, and genuine capability limitations.
Low expectations or legitimate assessment?
Scenario 1
A VP of Engineering inherits a team that the previous VP described as 'solid but not exceptional.' Over her first six months, she assigns the team's most challenging projects to a different group she considers 'high-calibre,' gives the inherited team maintenance work, and holds shorter, less frequent one-on-ones with them. At the six-month review, the inherited team's output metrics are 30% below the other group's.
Scenario 2
A startup CEO raises a Series B after strong growth. The lead investor tells the CEO privately: 'Your VP of Sales is not Series B calibre. You need to upgrade.' The CEO, who had previously been satisfied with the VP's performance, begins second-guessing the VP's pipeline calls, attends more of the VP's customer meetings, and starts interviewing replacement candidates. Over the next quarter, the VP's team misses their target for the first time.
Section 11
Top Resources
The Golem Effect sits at the intersection of social psychology, education research, and organisational behaviour. The strongest resources document the experimental evidence for expectation-driven performance, the transmission mechanisms, and the structural conditions that amplify or neutralise the effect.
The foundational study. Rosenthal and Jacobson demonstrated that teacher expectations — manipulated by fabricated test results — produced real IQ changes in students over an academic year. The book documents both the Pygmalion and Golem Effects, though the positive direction received more attention. The methodology has been debated and refined over five decades, but the core finding — that expectations transmit through behaviour and shape outcomes — has been replicated across cultures, age groups, and institutional settings.
Rosenthal's earlier work on experimenter expectancy effects, which provided the theoretical foundation for Pygmalion in the Classroom. The book documents how experimenters' expectations influenced the behaviour of laboratory rats — and by extension, any subject whose environment is shaped by the expectation-holder's behaviour. Essential for understanding the Golem Effect as a general principle of social interaction rather than a phenomenon specific to education.
Eden's Israeli Defense Forces study is the most rigorous military replication of the Pygmalion-Golem dynamic. By randomly assigning "high potential" labels to trainees, Eden demonstrated that instructor expectations — not trainee ability — drove performance differences. The study's controlled military context eliminates many confounds present in educational and corporate settings, making it one of the strongest pieces of evidence that expectation effects are real and consequential.
Eden's direct treatment of the Golem Effect as a productivity constraint in organisations. The paper argues that the Golem Effect is more common and more damaging than the Pygmalion Effect because low expectations are the default in most organisations — managers are more likely to assume mediocrity than excellence. Eden proposes structural interventions that neutralise the Golem channel without requiring managers to change their private assessments, focusing on standardised investment and blind evaluation.
Hastings and Meyer's account of Netflix's culture provides the most detailed case study of an organisation that structurally eliminated the conditions for the Golem Effect. The talent density model, the keeper test, and the radical candour framework all function as Golem defences — ensuring that every employee operates under high expectations with high investment, and that the low-expectation retention pattern that feeds the Golem loop does not persist. The book is not framed as a Golem Effect analysis, but the mechanisms it describes are precisely the structural interventions that the research literature recommends.
The Golem Effect — low expectations trigger systematic withdrawal of investment, which produces the underperformance the manager expected. The prophecy fulfils itself not because it was accurate but because it was acted upon.
Tension
Psychological Safety
Amy Edmondson's psychological safety is a direct structural countermeasure to the Golem Effect. In psychologically safe environments, team members can take interpersonal risks — asking questions, admitting mistakes, challenging assumptions — without fear of punishment. The Golem Effect thrives where psychological safety is low because the low-expectation individual cannot surface the dynamic: "I notice I'm getting less coaching than my peers" is a high-risk statement in an unsafe environment. Psychological safety gives the subject the voice to name the pattern, which is the first step in breaking the loop.
Reinforces
Talent Density
Talent density — the concept that team performance scales with the average capability in the room — interacts with the Golem Effect through differential investment. In low-talent-density environments, the Golem Effect is amplified: managers have low expectations of the team, invest less, and the team underperforms. In high-talent-density environments, the default expectation is high, which triggers the Pygmalion channel. Netflix's talent density strategy is an implicit Golem defence: by maintaining high expectations as the uniform standard, the organisation eliminates the low-expectation condition that the Golem Effect requires.
Tension
10x Individuals/Teams
The 10x concept — that exceptional individuals produce order-of-magnitude more output than average ones — interacts with the Golem Effect in a dangerous way. Organisations that believe in 10x performers tend to invest heavily in identifying and developing them. The risk: everyone not labelled 10x receives systematically less investment, triggering the Golem cascade. The 10x individual may be partly a Pygmalion creation — high expectations producing high performance. The "average" performer may be partly a Golem creation — low expectations producing low performance. The antidote is not to abandon the 10x concept but to ensure that investment differentials reflect demonstrated capability under equal conditions, not pre-emptive labelling. Assume competence until proven otherwise — then prove it with data from standardised investment.
The organisational fix is investment standardisation, not expectation training. You cannot train managers to stop having expectations. Expectations are automatic, pre-conscious, and deeply resistant to modification through awareness programmes. What you can do is build systems that hold investment constant regardless of the manager's private assessment. Standardised onboarding programmes. Equal distribution of stretch assignments. Mandatory coaching cadences. Blind evaluation of work product. These structures do not eliminate the Golem Effect — the climate channel still operates through micro-behaviours that no policy can regulate. But they neutralise the input, output, and feedback channels, which carry the majority of the effect's force.
The uncomfortable personal question: Who are you currently operating on with low expectations? Not the people you have formally assessed as underperformers and are actively managing. The people you have quietly written off — the direct report you have stopped developing, the team you have stopped investing in, the project you have stopped championing. Examine whether the evidence of their underperformance preceded or followed your withdrawal of investment. If the withdrawal came first, you are not observing a truth about their capability. You are observing the output of your own expectation. And you are calling it data.
Scenario 3
A product organisation has three teams: Platform (labelled 'strategic'), Growth (labelled 'important'), and Internal Tools (labelled 'support function'). Over two years, Platform receives 3x the engineering headcount, direct access to the CPO, and first priority on shared infrastructure. Internal Tools receives minimal headcount, quarterly CPO check-ins, and last priority on shared resources. Platform ships two major products. Internal Tools' engineer satisfaction scores drop 40% and three of five engineers leave. Leadership cites the attrition as evidence that Internal Tools 'can't retain talent.'
