Standing at the Edge: What the Glass Cliff Evidence Really Says and How Boards Can Set Crisis Leaders Up to Succeed
Download the article here:
Listen to a review of this article:
Abstract: The glass cliff describes a pattern in which women and members of racial and ethnic minority groups are more likely than white men to be appointed to leadership roles when an organization is already in trouble. Two decades after the term was coined, the idea is widely cited and sometimes dismissed, and a 2026 controversy at a global consulting firm brought the debate back into public view. This article reviews what the research shows, including a 2020 meta-analysis that found small, context-dependent effects, recent large archival studies that point in opposite directions, and evidence that women and minority leaders tend to receive less leeway once appointed. It then examines documented appointments across automotive, banking, technology, media, sports, and legal settings, and sets out five evidence-based responses for boards: risk-transparent appointments, early and inclusive succession pipelines, resourced transitions, fair performance attribution, and evidence-first governance of contested debates. It closes with three pillars for building durable succession capability.
In late September 2026, a LinkedIn exchange about a new book on women's leadership turned into a reputational problem for one of the world's best-known consulting firms. A senior McKinsey partner, who also held a director role at the McKinsey Global Institute, questioned in a series of comments whether the "glass cliff" exists at all, suggested the idea offered a convenient way to claim victimhood, and said he would not spend time researching it (Boyle et al., 2026; Mohan, 2026). The firm said it was disappointed, that the remarks were made in a personal capacity, and that it did not endorse them (Boyle et al., 2026). After an internal review, McKinsey confirmed on October 11 that it had removed him from his leadership roles, with a spokesperson saying the comments "fell well short of the standards we expect" (Pereira, 2026). He remained a senior partner, and the posts were deleted (Pereira, 2026).
The episode drew reactions in both directions. Some commentators read it as a firm holding a senior leader accountable for dismissing a body of research that the firm's own publications had drawn on; others saw a professional being sanctioned for voicing skepticism about a contested academic idea. Both readings are worth taking seriously, because both point at the same underlying question that boards, search committees, and senior leaders face every time a top job opens in difficult circumstances: what does the evidence actually say about who gets handed the hardest leadership assignments, and what should organizations do about it?
That question matters well beyond one firm or one comment thread. Women now lead a record 56 companies on the Fortune 500, or 11.2% of the list (Hinchliffe, 2026). Data reported by Bloomberg suggest that, over the past decade, women CEOs of Russell 3000 companies left after an average tenure of 6.2 years compared with 8.9 years for men, and were more likely to be forced out (Boyle et al., 2026). Whether that gap reflects precarious appointments, harsher evaluation, different career choices, or some mix of these is exactly what the glass cliff literature tries to untangle.
This article does three things. First, it sets out what the research does and does not show, including findings that challenge the original claim. Second, it looks at documented appointments in several industries where commentators and researchers have raised glass cliff questions, without assuming anyone's motives. Third, and most importantly, it lays out what boards and leaders can do so that any leader taking on a turnaround, regardless of gender or background, is set up to succeed rather than set up to be blamed. The practical case does not depend on settling every academic dispute. Even skeptics of the glass cliff should want appointment processes that disclose real risks, start building pipelines before a crisis hits, resource new leaders properly, and evaluate them fairly.
The Glass Cliff Landscape
Defining Key Terms
The glass ceiling refers to the invisible barriers that keep women and minority professionals from reaching senior roles. The glass cliff, a term introduced by Michelle Ryan and Alex Haslam at the University of Exeter, describes what can happen to those who do break through: they are more likely than their white male peers to be appointed to leadership positions that are precarious, meaning roles associated with a higher risk of failure and criticism because the organization is already struggling (Ryan & Haslam, 2005). The metaphor is about position, not ability. A leader standing at the edge of a cliff is not less capable; the ground under them is simply less stable.
Three related concepts recur in the research. The first is think crisis, think female, the finding that the classic association between being a manager and having stereotypically masculine traits weakens or reverses when people imagine an organization in trouble, so that stereotypically feminine traits become more strongly linked to what a leader in a failing company should be like (Ryan et al., 2011). The second is the savior effect, a term coined by Alison Cook and Christy Glass to describe what happens after a nontraditional leader's tenure: when performance declines under a woman or minority CEO, that leader is likely to be replaced by a white man, as if order is being restored (Cook & Glass, 2014a). The third is signaling, the idea that boards may appoint a nontraditional leader partly to send a visible message of change to investors, employees, and the public (Bruckmüller & Branscombe, 2010; Reinwald et al., 2023).
It helps to be clear about what the glass cliff claim is not. It is not a claim that every woman or minority executive is appointed to a failing company, nor that every difficult appointment is a trap, nor that boards consciously set leaders up to fail. Researchers who developed the concept have said explicitly that glass cliffs are neither universal nor inevitable (Ryan & Haslam, 2009). The claim is probabilistic and contextual: on average, and under certain conditions, crisis increases the chance that a nontraditional candidate is chosen, and the circumstances surrounding that appointment can make success harder to achieve and harder to recognize.
A precarious context precedes the appointment; the new leader receives limited support or incomplete information; inherited problems are attributed to the leader; the leader is replaced, often by someone with a more traditional profile; and beliefs about who can lead in hard times are quietly reinforced, setting up the next crisis appointment. Each stage is also a point at which a board can intervene, which is the logic behind the responses discussed later in this article.
Prevalence, Drivers, and Distribution
The original evidence. Ryan and Haslam (2005) examined FTSE 100 companies around the time they appointed men or women to their boards. Using share price as the measure of performance, they found that companies appointing women had experienced consistently poor share price performance in the months before the appointment, while those appointing men had relatively stable performance. Performance did not drop after women joined, which suggested that the women had been appointed into difficulty rather than causing it (Ryan et al., 2016).
The first major challenge. Adams et al. (2009) looked at U.S. CEO appointments and found no evidence that women were appointed to firms with worse prior accounting performance, concluding that the pattern did not appear among female CEOs at U.S. firms. Ryan and Haslam (2009) replied that glass cliffs are not universal and that a null result on one set of financial measures should not be over-read. Their later review argued that part of the disagreement reflects how performance is measured: stock-based measures capture investor sentiment, while accounting measures such as return on assets capture operating results, and glass cliff patterns have tended to show up more clearly in the former (Ryan et al., 2016).
Experimental evidence. Laboratory and scenario studies have been more consistent. In three experiments with management graduates, high school students, and business leaders, participants were more likely to choose an equally qualified woman over a man when an organization's performance was declining rather than improving (Haslam & Ryan, 2008). Bruckmüller and Branscombe (2010) found that the preference for women in crisis appeared mainly when the organization had a history of male leadership, consistent with a desire to signal change. Similar patterns have appeared in politics, where women in the United Kingdom's Conservative Party contested harder-to-win seats than men in the 2005 general election (Ryan et al., 2010), and ethnic minority candidates for the same party faced opponents with larger majorities across three elections (Kulich et al., 2014).
The meta-analysis. The most comprehensive synthesis to date pooled 74 studies containing 91 independent samples (Morgenroth et al., 2020). Its findings are easy to overstate in either direction, so they are worth stating carefully. Across 28 archival field samples on gender, the average glass cliff effect was small (Hedges' g = 0.07) and varied substantially by domain: it held in politics, education, and nonprofit settings but not, on average, in corporate management or sports. In experiments where each participant evaluated only one candidate, women were rated as more suitable for crisis leadership than men (g = 0.32), but in experiments where participants compared men and women side by side, the effect was small and not statistically significant. In leader-selection experiments, women were more likely to be chosen in crisis than in non-crisis conditions (odds ratio = 1.45 across 35 samples), and this effect was larger in countries with greater gender inequality. The glass cliff also extended to racial and ethnic minority groups, with a small archival effect (g = 0.21) that did not differ statistically from the gender estimate (Morgenroth et al., 2020).
Recent large-sample studies point in different directions. Reinwald et al. (2023) analyzed 26,156 executive appointments in U.S. firms from 2000 to 2016, using a regression discontinuity design around a financial distress threshold to get closer to a causal estimate. They found that crisis status significantly increased the likelihood of appointing a woman to the top management team, that crisis firms were more likely to describe female appointments in change-oriented language in press releases, and that the effect depended on context: it was concentrated in firms with no other female executive and those under greater investor attention. By contrast, Li et al. (2025), studying more than 10,000 U.S. CEO appointments between 1998 and 2022, of which 526 were women, reported no clear evidence that women were more likely to be appointed CEO at financially distressed firms; where patterns appeared, women were more likely to be appointed at financially healthier companies (as summarized in secondary coverage of the study). Cook and Glass's own work is similarly nuanced. Their analysis of Fortune 500 CEO transitions found that white women and people of color were more likely than white men to be promoted to CEO at weakly performing firms (Cook & Glass, 2014a), while a separate analysis focused on women found that board diversity, rather than prior firm performance, best predicted women's promotion and tenure (Cook & Glass, 2014b).
What drives the pattern where it exists. The research points to several mechanisms rather than one. Stereotypes play a role, since people associate stereotypically communal traits with managing people through hardship (Ryan et al., 2011). Signaling plays a role, since the effect is stronger where a new appointment would visibly break with a male-led past (Bruckmüller & Branscombe, 2010; Reinwald et al., 2023). Access to options plays a role, since leaders who have had fewer opportunities may weigh a risky role differently, although women in experimental research generally rated risky positions as less attractive than men did (Ryan et al., 2016). And the availability of stakeholder support matters: the preference for women in crisis was strongest when the new leader could not count on full stakeholder backing (Rink et al., 2013).
Distribution. Representation itself remains thin, which makes every appointment more visible. Women's share of Fortune 500 CEO roles rose from two companies in 1998 to 41 in 2021, 52 in 2023, and 56 in 2026, or 11.2% of the list (Hinchliffe, 2022, 2026; Hinchliffe & Abrams, 2024; Ajemian, 2025). In the United Kingdom, women held 43% of FTSE 350 board seats in 2026 but only 8% of CEO roles and 17% of chair roles (Freshfields, 2026). When a group holds roughly one in ten top jobs, a handful of high-profile exits can shape perceptions far more than they would for the majority group.
The fair summary is this. The glass cliff is real in the sense that it shows up reliably in experiments and in a meta-analysis of field data, especially for racial and ethnic minorities and in some sectors. It is small and contingent in corporate archival data, and at least one large recent study of U.S. CEOs did not find it. People who say the phenomenon is a myth are overstating what the evidence shows, and so are people who describe it as a universal rule. For boards, the more useful question is not whether the effect exists on average across all firms, but whether their own appointment practices create the conditions in which it occurs.
Organizational and Individual Consequences
Organizational Performance Impacts
The organizational costs of glass cliff dynamics come less from the appointment itself than from what follows.
Churn and discontinuity. CEO turnover is expensive in any form, and turnover driven by misattribution is doubly so. If a board replaces a leader for failing to reverse problems that were structurally entrenched before they arrived, the organization pays for two transitions and may still not have addressed the root causes. Cook and Glass (2014a) found that when performance declined during a woman or minority CEO's tenure, the most likely successor was a white man. Ryan et al. (2016) interpreted this as part of a pattern in which nontraditional leaders receive less leeway, and in which the return of a conventional leader is treated as a return to normal.
Distorted learning. When a turnaround fails, boards need to understand why. Was the strategy wrong, the execution weak, the resources insufficient, or the situation unrecoverable? If attribution defaults to the leader, the organization learns the wrong lesson. Kulich et al. (2007) found in an experiment with 210 participants that performance-based bonuses for male leaders tracked company performance through perceived charisma, while bonuses for female leaders were less directly tied to company performance. If evaluators disconnect a leader's rewards from the results they produce, they also disconnect the organization's learning from what actually happened.
Contested performance effects of diversity mandates. Boards also need to be careful not to read too much into studies of what happens when the composition of leadership changes. Norway's 2003 law requiring 40% representation of each sex on public limited company boards is the most studied example. Ahern and Dittmar (2012) reported that share prices fell when the law was announced and that Tobin's Q (a market-based valuation ratio) declined in the following years, with boards becoming younger and less experienced. Eckbo et al. (2022) revisited the same episode and concluded that, once problems with dating the news and correlated stock returns were addressed, the valuation effect was statistically insignificant. Matsa and Miller (2013) found that affected firms undertook fewer workforce reductions, raising labor costs and lowering short-term profits. The lesson for glass cliff debates is that performance effects depend heavily on the measure, the comparison group, and the time horizon, which is precisely why fair attribution matters.
Signaling value and its limits. Reinwald et al. (2023) found that crisis firms used more change-related language when announcing female appointments. A visible appointment can buy goodwill with investors and employees, but a signal is not a strategy. If the appointment is used to communicate change without the authority, resources, and time to deliver it, the signal can turn into a liability when results lag.
Individual Wellbeing and Stakeholder Impacts
Scrutiny and dismissal risk. Several studies find that women CEOs face more external pressure. Gupta et al. (2018) found that female CEOs were significantly more likely than male CEOs to be targeted by activist investors. Gupta et al. (2020) found that, holding other factors constant, female CEOs were more likely to be dismissed, and that while male CEOs were less likely to be dismissed when firm performance was high, female CEOs faced similar dismissal likelihood regardless of performance. That second finding goes to the heart of the glass cliff concern: if strong results do not protect a leader, the incentive to accept a difficult role weakens.
Support, authority, and tenure. Glass and Cook (2016), drawing on the career histories of women who had served as Fortune 500 CEOs and interviews with women executives, found that women were more likely to be promoted into high-risk roles, often lacked the support or decision-making authority they needed, and tended to have shorter tenures. In experimental work, participants who endorsed glass cliff appointments also tended to believe such roles would be especially stressful for women (Haslam & Ryan, 2008). The individual cost is not just a shortened tenure; it is a career narrative that may follow the leader into every subsequent search.
Public exposure. Leaders of large or highly visible organizations now face scrutiny that plays out in real time on social platforms. The 2015 episode at Reddit, discussed below, involved a petition calling for the interim CEO's resignation that gathered more than 190,000 signatures (Mohan, 2015). Whatever one's view of the underlying decisions, the speed and personal tone of that kind of pressure is a wellbeing risk that boards should plan for when they appoint someone into a volatile situation.
Stakeholders and the pipeline. The consequences extend beyond the individual. Employees, especially those from underrepresented groups, watch how the first women or minority leaders in their organizations are treated. If those leaders are appointed in crisis and removed quickly, the message received may be that top roles are available only when no one else wants them. Some researchers also worry about the opposite problem: Li et al. (2025) argued that overstating the glass cliff could itself discourage women from seeking senior roles, according to secondary coverage. Both concerns point toward the same remedy, which is accurate information about the evidence and about each specific role.
Evidence-Based Organizational Responses
The research does not support a single fix, but it does suggest where the leverage lies. Glass cliff dynamics are most likely when crisis appointments are made quickly, from narrow candidate pools, with incomplete information, little stakeholder backing, and evaluation standards that do not separate inherited conditions from leader performance. Table 1 summarizes five responses that address those conditions, the evidence behind each, and organizations whose documented experience illustrates them.
Table 1: Evidence-Based Responses to Glass Cliff Dynamics in Leadership Appointments
Response | Evidence Base | Key Practices | Illustrative Examples |
Risk-transparent appointments | Experimental and archival evidence that crisis shapes who is chosen (Haslam & Ryan, 2008; Morgenroth et al., 2020) | Candidate due diligence pack; disclosure of known liabilities; explicit mandate and success criteria | GM and the ignition-switch recall; Citigroup and its 2020 consent orders; Xerox in 2001 |
Structured, inclusive succession pipelines started early | Decision-maker diversity predicts women's promotion and tenure (Cook & Glass, 2014b); Rooney Rule study (DuBois, 2016) | Multi-year succession planning; diverse slates for every senior role; transparent criteria | NFL Rooney Rule; Mansfield Rule at Faegre Drinker; EU Directive 2022/2381 selection rules |
Resourcing and backing turnaround leaders | Stakeholder support attenuates the glass cliff (Rink et al., 2013); lack of authority shortens tenure (Glass & Cook, 2016) | Visible board sponsorship; agreed resources and authority; transition coaching; realistic time horizon | Xerox under Mulcahy; Reddit's 2015 interim leadership; X under Yaccarino |
Fair performance attribution and evaluation | Gendered attribution of results (Kulich et al., 2007); dismissal asymmetry (Gupta et al., 2020); savior effect (Cook & Glass, 2014a) | Baseline at appointment; separate legacy and new issues; staged reviews; succession audits | Citigroup's regulatory remediation; Yahoo under Mayer |
Accountability and evidence-first voice | Contested and mixed findings (Li et al., 2025; Reinwald et al., 2023); quota debates (Ahern & Dittmar, 2012; Eckbo et al., 2022) | Distinguish conduct from viewpoint; cite the full evidence; review processes for contested topics | McKinsey's 2026 conduct review; Nasdaq board diversity rule litigation |
Risk-Transparent Appointments: Due Diligence for the Appointee
Boards conduct exhaustive due diligence on candidates. They rarely provide candidates with equivalent due diligence on the role. Yet the glass cliff literature suggests that the information gap between the appointing body and the appointee is one of the mechanisms through which precarious appointments become traps. Experimental studies show that crisis context changes who is selected (Haslam & Ryan, 2008; Bruckmüller & Branscombe, 2010), and the meta-analysis found the selection effect was stronger when the gender of the previous leader was unknown, implying that how the situation is framed matters (Morgenroth et al., 2020). If candidates do not understand what they are walking into, they cannot negotiate the resources, authority, or timeline the situation requires.
Effective approaches:
Appointee due diligence pack: Provide finalists with a confidential briefing that covers known legal, regulatory, financial, operational, and cultural liabilities, including open investigations and enforcement actions.
Explicit mandate: Write down what the board is asking the leader to achieve, by when, and what "success" looks like at 6, 18, and 36 months, distinguishing stabilization from growth.
Independent view of the baseline: Commission an independent assessment of the organization's condition at the point of appointment so that later evaluation has a fixed reference point.
Negotiation space: Invite candidates to set conditions on resources, team changes, and authority before accepting, and treat those requests as signs of judgment rather than reluctance.
Symmetric scrutiny: Ask whether the board would describe the role in the same way to every candidate on the slate, and record that it has done so.
General Motors offers a widely discussed illustration of how much can be inherited. Mary Barra became GM's first female CEO in January 2014, and within weeks the company was managing a recall connected to faulty ignition switches on models dating back more than a decade, which had been linked to at least 13 deaths (CBS News, 2014; Watkins & Duke, 2014). She led the public response, including congressional testimony, and the company commissioned former U.S. attorney Anton Valukas to investigate. His report found that GM had failed for more than a decade to address the defect and described a culture marked by a proliferation of committees and a lack of accountability, while finding no evidence of a cover-up (CBS News, 2014). Fifteen employees were dismissed. Commentators at the time cited Barra's situation as a possible glass cliff example (Eichler, 2016). What is clear from the documented timeline is that the scale of the legacy problem did not become public until after the appointment, and that the independent investigation helped separate the history of the defect from the new leadership's response.
Citigroup illustrates a case in which the problem was public before the appointment. In 2020, the Federal Reserve and the Office of the Comptroller of the Currency took enforcement action requiring Citigroup to address weaknesses in risk management, data governance, and internal controls. Jane Fraser became CEO in March 2021, inheriting that remediation. In July 2024, the two regulators fined the bank a combined $135.6 million for insufficient progress, particularly on data quality management, while the OCC's acting head noted that meaningful progress had been made overall (Federal Reserve Board, 2024; Hollerith, 2024). Because the consent orders predated her tenure and were publicly documented, the terms of the challenge were visible to anyone evaluating her performance, which is exactly the kind of baseline that risk transparency is meant to create.
A common pitfall is treating disclosure as a liability exercise rather than a leadership one. Boards that share only what legal counsel requires, or that present a difficult situation optimistically to land a preferred candidate, may secure the appointment but undermine the relationship and the later evaluation. Over-disclosure is rarely the problem in crisis appointments.
Structured, Inclusive Succession Pipelines Started Early
If underrepresented candidates are considered seriously only when an organization is already in trouble, the glass cliff is almost guaranteed to appear in the data. The most robust structural response is to make diverse candidates visible and prepared for senior roles long before any crisis. Cook and Glass (2014b) found that diversity among board decision makers, not firm performance, significantly increased women's chances of being promoted to CEO and lengthened their tenure regardless of how the firm performed. Their university's summary of this research program framed the practical implications as formalizing the hiring process, making it transparent, and pushing decision makers beyond personal networks (Utah State University, 2013).
Effective approaches:
Continuous succession planning: Maintain a living succession plan for the CEO and each top-team role, reviewed at least annually, with ready-now, ready-soon, and longer-term candidates.
Diverse slates by default: Require that interview slates for senior roles include candidates from underrepresented groups in good times as well as bad, so that crisis is not the only entry point.
Development before vacancy: Give high-potential leaders profit-and-loss responsibility, board exposure, and stretch assignments years before they are needed.
Transparent criteria: Define selection criteria in advance, apply them consistently, and record how each candidate was assessed against them.
Board composition: Treat the diversity of the nominating committee and board as part of succession infrastructure, given evidence that decision-maker diversity shapes who gets chosen.
The National Football League's Rooney Rule, introduced in 2003, requires teams to interview at least one minority candidate for head coaching vacancies, without requiring that any particular candidate be hired. DuBois (2016) compared NFL hiring before and after the rule with NCAA head coaches and NFL coordinators as comparison groups and estimated that a minority candidate was roughly 19% to 21% more likely to fill an NFL head coaching vacancy after the rule than before. The sports setting is relevant here because Cook and Glass (2013) had earlier found that minority coaches in college basketball were more likely to be hired into losing programs and had shorter tenures, a field example of the glass cliff and savior dynamics in action.
Faegre Drinker, a U.S. law firm, has participated in Diversity Lab's Mansfield Rule since its launch in 2017. In the version the firm reported on in 2022, certification required considering at least 30% women, underrepresented racial and ethnic lawyers, lawyers with disabilities, and LGBTQ+ lawyers for top leadership roles, senior lateral hires, and promotions into equity partnership, and tracking candidate pools in a disaggregated way. The firm reported being one of 108 firms to earn the program's "Plus" designation for reaching at least 30% representation in a notable number of leadership roles (Faegre Drinker, 2022). The evidence on the rule's outcomes is mixed: a working paper comparing certified and non-certified firms from 2007 to 2019 found no statistically significant effect on diversity growth, promotion, or committee composition (Cecchi-Dimeglio, 2021). That finding is a useful caution that slate requirements work best as part of a broader pipeline system rather than as a standalone fix.
The European Union has taken a regulatory route. Directive (EU) 2022/2381 requires large listed companies to reach either 40% of non-executive director roles or 33% of all director roles held by the underrepresented sex by June 30, 2026. Companies that fall short must use selection procedures based on clear, neutrally formulated criteria, give priority to the underrepresented sex where candidates are equally qualified, and share selection criteria with unsuccessful candidates on request (McCann FitzGerald, 2025).
A common pitfall is building a diverse slate only for the urgent vacancy. If the pipeline work has not been done, the diverse candidate may be less prepared, less known to the board, and more likely to be appointed in precisely the circumstances where support is thinnest.
Resourcing and Backing Leaders in Turnaround Roles
The single most actionable finding in the experimental literature may be that stakeholder support changes the glass cliff. Rink et al. (2013) showed that the preference for women in crisis was reduced when stakeholders supported the decision to appoint a new leader and persisted when that support was absent. Glass and Cook (2016) found that women CEOs frequently lacked the institutional support and decision-making authority to carry out their strategies, which contributed to shorter tenures. Taken together, these findings suggest that the question is not only who is appointed but what they are given.
Effective approaches:
Visible board sponsorship: Assign a lead director or chair to act as the new leader's sponsor, meet regularly, and publicly back the mandate during the early, most volatile period.
Agreed resources: Settle the budget, capital, and talent the leader can draw on before announcing the appointment, and revisit them at fixed intervals.
Real authority: Clarify decision rights, especially where there is a controlling shareholder, founder, or parent company, so that accountability matches authority.
Transition support: Offer executive coaching, a structured onboarding plan, and access to former turnaround leaders as peer advisors.
Realistic time horizon: Agree how long a turnaround of this kind typically takes and protect that window from short-term pressure where possible.
Wellbeing and security planning: For highly visible roles, plan for online harassment and personal security as part of the support package.
Xerox under Anne Mulcahy is often cited as a turnaround in which the leader had both the mandate and the room to act. When she became CEO in 2001, the company carried more than $17 billion in debt, had lost money in each of the previous six years, faced an SEC investigation into accounting at its Mexico unit, and was close to a bankruptcy filing; the share price fell 15% on the day her appointment was announced (Vollmer, 2004). She raised $2.5 billion in cash, cut capital expenditure and overhead sharply while protecting research and development, spent her first 90 days listening to employees and customers, and returned the company to profitability (Vollmer, 2004). The case is a reminder that precarious appointments can succeed, and that the conditions of authority and time matter.
Reddit's experience in 2015 shows how quickly support can erode. Ellen Pao was serving as interim CEO when the company dismissed a well-liked employee who coordinated its popular "Ask Me Anything" forum; volunteer moderators temporarily shut down nearly 300 communities in protest, and a petition calling for Pao's resignation passed 190,000 signatures (Mohan, 2015). She apologized publicly, acknowledged the company had made mistakes, and resigned on July 10, 2015, with co-founder Steve Huffman returning as CEO (Brooks, 2015). The documented record does not tell us how decisions were divided inside the company. It does show an interim leader carrying public accountability in a volatile community with a short and uncertain mandate.
X offers a more recent example of the importance of decision rights. Linda Yaccarino was appointed CEO in 2023 with a remit to help rebuild the platform's relationships with advertisers, many of whom had pulled back over content concerns. She announced her resignation in July 2025, a day after the company's affiliated AI chatbot posted antisemitic content, and months after X had been acquired by xAI; she did not give a specific reason, and one analyst quoted in coverage suggested her approach had clashed with that of the owner (The Globe and Mail, 2025). Whatever the reasons for her departure, the case illustrates the general point that a leader cannot be fully accountable for outcomes that depend on decisions made elsewhere in the ownership structure.
A common pitfall is confusing a public statement of confidence with actual support. Announcing that the board is "fully behind" a new CEO costs nothing. Agreeing on the resources, the decision rights, and the time horizon in advance, and honoring them when the first quarter disappoints, is what the research suggests makes the difference.
Fair Performance Attribution and Evaluation
Attribution is where glass cliff dynamics become self-reinforcing. If evaluators credit or blame leaders differently depending on who they are, then the same results can produce different careers. Kulich et al. (2007) found that bonus allocation for male leaders followed company performance, while for female leaders it was less tightly linked to results. Gupta et al. (2020) found that strong firm performance reduced dismissal risk for male CEOs but not for female CEOs. Cook and Glass (2014a) documented the savior effect, in which declining performance under a nontraditional CEO tends to be followed by appointment of a white man. Ryan and Haslam (2005) originally emphasized that performance problems preceded women's appointments rather than following them, which is the essential distinction that any fair evaluation process must preserve.
Effective approaches:
Fixed baseline: Record the organization's condition at the date of appointment using a pre-agreed set of indicators, and evaluate change from that baseline rather than absolute levels alone.
Legacy and new issues tracked separately: Maintain a register of problems inherited at appointment and report progress on them separately from new initiatives.
Multiple measures: Look at market, accounting, operational, and cultural indicators together, given evidence that glass cliff findings depend on which performance measure is used (Ryan et al., 2016).
Staged reviews: Use structured reviews at agreed milestones rather than reacting to a single quarter, and document the reasoning for any leadership change.
Succession audits: Periodically review the board's own record of appointments and exits by gender and race, including the condition of the organization at each point, to detect savior patterns.
Citigroup's regulatory record shows why separating inherited issues matters in practice. The 2020 enforcement actions set the remediation agenda that the next CEO inherited, and the 2024 penalties were explicitly framed by the Federal Reserve as a consequence of insufficient progress on data quality under the 2020 action (Federal Reserve Board, 2024). A fair evaluation would ask both whether progress has been fast enough and how difficult the inherited problem was, rather than treating the existence of the fine alone as evidence about leadership quality.
Yahoo under Marissa Mayer is frequently cited in commentary about the glass cliff. She joined in 2012 after a period of high CEO turnover at the company, oversaw some growth in mobile, and also oversaw the disclosure of massive data breaches. When Verizon completed its purchase of Yahoo's core business in 2017, the price had been cut from $4.8 billion to $4.5 billion after those disclosures (Silverman, 2017). A newspaper column in 2016 suggested she may have fallen from a glass cliff (Eichler, 2016). Assessing her tenure fairly requires separating the strategic position Yahoo was in when she arrived from the decisions made during her time in office, which is exactly the analysis that baselines and legacy-issue registers are designed to support.
A common pitfall is assuming that a fast exit settles the question of performance. A quick replacement can feel decisive, but if the root causes were inherited, the next leader starts from the same place, and the organization loses both time and credibility.
Accountability and Evidence-First Voice
The McKinsey episode raises a different kind of question: how should organizations handle public disagreement about contested research on issues such as gender and leadership? There are two ways to get this wrong. One is to treat any skepticism about a research finding as misconduct, which discourages the kind of debate through which science actually improves. The other is to treat dismissive commentary by senior leaders as a purely personal matter, when it can shape how colleagues and clients understand the organization's commitments. The glass cliff literature itself models a better path, with critics and originators publishing responses to one another (Adams et al., 2009; Ryan & Haslam, 2009), and with new large-sample studies continuing to test and refine the claim (Reinwald et al., 2023; Li et al., 2025).
Effective approaches:
Distinguish conduct from viewpoint: Evaluate senior leaders' public statements on tone, accuracy, and respect for colleagues, not on whether they reach a particular conclusion.
Cite the full evidence: When the organization communicates about contested research, present the strongest findings on each side, including null and mixed results.
Clear process: Use a defined, documented process for reviewing public comments by senior leaders, with outcomes proportionate to the conduct.
Internal literacy: Equip directors and executives with a short, balanced evidence briefing on topics, such as the glass cliff, that affect appointment decisions.
Separate policy from proof: Recognize that the case for fair appointment practices does not depend on any single study, so that a contested finding need not derail sensible governance.
McKinsey's response unfolded in two stages. The firm first said it was disappointed by the partner's comments, did not endorse them, and stood by its research; it then conducted an internal review and removed him from his leadership roles, while he remained a partner (Boyle et al., 2026; Pereira, 2026). Commentators noted that the firm's own publications, including its long-running research with LeanIn, had addressed gender inequality and that a 2020 healthcare report had cited the glass cliff (Mohan, 2026). The firm's explanation focused on the comments falling short of its standards rather than on the scientific question, which is a meaningful distinction: an organization can hold leaders accountable for how they engage without adjudicating every academic dispute.
The fate of Nasdaq's board diversity rule shows how quickly the policy landscape can shift. The rule required Nasdaq-listed companies to disclose board diversity statistics and to have, or explain why they did not have, directors who self-identified as female, an underrepresented minority, or LGBTQ+. On December 11, 2024, the full Fifth Circuit Court of Appeals held in Alliance for Fair Board Recruitment v. SEC that the SEC had exceeded its authority in approving the rule (Paul Weiss, 2024). For boards, the lesson is that appointment practices grounded in sound governance, such as risk transparency, early pipelines, and fair evaluation, are more durable than practices that depend on a particular disclosure mandate.
A common pitfall is letting the loudest voices in a public argument set the organization's understanding of the evidence. Boards that rely on headlines, whether celebratory or dismissive, are likely to overreact in one direction or the other.
Building Long-Term Leadership Succession Capability
The five responses above are most effective when they are embedded in an organization's governance rather than revived each time a crisis arrives. Three pillars support that kind of durable capability.
Pillar 1: Succession as a Continuous Governance Discipline
The most reliable way to avoid glass cliff dynamics is to ensure that no one is considered for a top role for the first time in a crisis. That requires a succession process that runs every year, not only when a vacancy appears. The UK's FTSE Women Leaders Review has moved in this direction: after the FTSE 350 reached 43% women on boards, its 2026 report urged companies to move beyond "one and done" appointments and to reduce bias in senior selection, noting that women held only 8% of FTSE 350 CEO roles (Freshfields, 2026). Norway's experience suggests that the supply of qualified candidates expands when demand is clear: women appointed to boards after the quota were, on many measures, more qualified than those they succeeded, and the gender earnings gap within boards narrowed substantially (Bertrand et al., 2019). At the same time, Bertrand et al. (2019) found little evidence that the quota benefited women further down the organization, a reminder that board composition alone does not build an executive pipeline. A continuous succession discipline means mapping successors for every critical role, investing in their development years ahead, and making sure the board knows them well before they are needed.
Pillar 2: Measurement, Baselines, and Learning Systems
Organizations cannot manage what they do not measure. Boards should keep a simple record of every senior appointment that captures the organization's condition at the time, the support and mandate provided, the length of tenure, the reason for exit, and the profile of the successor. Over time, that record lets a board see whether its own decisions show a savior pattern, whether certain roles are consistently precarious, and whether different leaders received different levels of support. The research shows why this is worth doing: findings vary substantially by sector, measure, and country (Morgenroth et al., 2020), so the only way to know whether glass cliff dynamics affect a particular organization is to look at its own data. Pairing that record with fixed baselines at appointment, as described above, turns evaluation from a judgment about a person into an analysis of a situation and a response.
Pillar 3: A Culture of Evidence-First Debate and Visible Backing
The third pillar is cultural. Leaders who take on difficult roles need to know that the board will stand behind a reasonable strategy through a rough patch, and that their results will be judged on the same terms as anyone else's. Organizations also need norms for discussing contested topics such as the glass cliff that welcome evidence on all sides without tolerating dismissiveness toward colleagues. Building that culture includes training directors to recognize the attribution biases documented in the literature (Kulich et al., 2007; Gupta et al., 2020), rewarding executives who develop diverse successors, and making clear that taking on a hard assignment is a career asset rather than a career risk. When a turnaround does not work, the organization should be able to say clearly why, without defaulting to the conclusion that the wrong kind of person was in charge.
Conclusion
The glass cliff has become a shorthand that people use to mean very different things. In its careful form, it is a modest and testable claim: under some conditions, crisis increases the likelihood that women and minority candidates are chosen for leadership, and the circumstances of those appointments can make success harder to achieve and recognize. The evidence for that claim is strongest in experiments, in some public and nonprofit settings, and for racial and ethnic minorities; it is weaker and contested in corporate archival data, where at least one large recent study found no effect for women CEOs (Morgenroth et al., 2020; Li et al., 2025). Evidence on what happens after appointment, including greater dismissal risk regardless of performance, more activist pressure, and replacement by traditional profiles when results disappoint, is in some ways more concerning than the evidence on selection itself (Cook & Glass, 2014a; Gupta et al., 2018, 2020).
Boards do not need to resolve every academic debate to act. The practices that reduce glass cliff risks are the same practices that improve leadership transitions for everyone: telling candidates the truth about the role, developing a diverse bench before a crisis, giving new leaders the authority, resources, and time they need, evaluating them against an honest baseline, and handling contested debates with evidence and respect. Each organization in this article, from GM and Citigroup to Xerox, Reddit, Yahoo, and X, faced a version of the same question about what a leader inherits and what they can fairly be held accountable for. The organizations that answer it well will get more out of every leader they appoint.
Practical takeaways for boards and senior leaders:
Give candidates due diligence: Share a full, confidential picture of known risks and liabilities with every finalist, and agree on a written mandate before the appointment.
Build the bench early: Maintain multi-year succession plans with diverse candidates for every critical role so that crisis is never the first time someone is considered.
Back the appointment with resources and authority: Assign a board sponsor, settle resources and decision rights, provide transition coaching, and agree on a realistic time horizon.
Evaluate against a baseline: Record the organization's condition at appointment, separate inherited issues from new ones, and use multiple performance measures.
Audit your own pattern: Track appointments, tenures, exits, and successors by gender and race to detect any savior effect in your own decisions.
Debate with evidence: Present the full range of research on contested topics and judge leaders' public comments on conduct and accuracy, not on conclusions alone.
Research Infographic
References
Adams, S. M., Gupta, A., & Leeth, J. D. (2009). Are female executives over-represented in precarious leadership positions? British Journal of Management, 20(1), 1–12.
Ahern, K. R., & Dittmar, A. K. (2012). The changing of the boards: The impact on firm valuation of mandated female board representation. The Quarterly Journal of Economics, 127(1), 137–197.
Ajemian, N. (2025, June 2). Women run 11% of Fortune 500 companies in 2025, but progress is still slow. Fortune.
Bertrand, M., Black, S. E., Jensen, S., & Lleras-Muney, A. (2019). Breaking the glass ceiling? The effect of board quotas on female labour market outcomes in Norway. The Review of Economic Studies, 86(1), 191–239.
Boyle, M., Choudhury, A., & Bainbridge, A. (2026, September 30). McKinsey partner's online comments on women CEOs spark rebuke. Insurance Journal (Bloomberg).
Brooks, J. (2015, July 10). Reddit's Ellen Pao: "We screwed up." KQED.
Bruckmüller, S., & Branscombe, N. R. (2010). The glass cliff: When and why women are selected as leaders in crisis contexts. British Journal of Social Psychology, 49(3), 433–451.
CBS News. (2014, June 5). GM chief says no cover-up in delayed recall.
Cecchi-Dimeglio, P. (2021). The impact of "soft" affirmative action policies on women and minority demographics, hiring, promotion, attrition, and leadership in law firms: The "Mansfield Rule" [Working paper]. SSRN.
Cook, A., & Glass, C. (2013). Glass cliffs and organizational saviors: Barriers to minority leadership in work organizations? Social Problems, 60(2), 168–187.
Cook, A., & Glass, C. (2014a). Above the glass ceiling: When are women and racial/ethnic minorities promoted to CEO? Strategic Management Journal, 35(7), 1080–1089.
Cook, A., & Glass, C. (2014b). Women and top leadership positions: Towards an institutional analysis. Gender, Work & Organization, 21(1), 91–103.
DuBois, C. (2016). The impact of "soft" affirmative action policies on minority hiring in executive leadership: The case of the NFL's Rooney Rule. American Law and Economics Review, 18(1), 208–233.
Eckbo, B. E., Nygaard, K., & Thorburn, K. S. (2022). Valuation effects of Norway's board gender-quota law revisited. Management Science, 68(6), 4112–4134.
Eichler, L. (2016). Female CEOs may find themselves falling off a glass cliff. The Globe and Mail.
Faegre Drinker. (2022). Faegre Drinker achieves Mansfield 5.0 Certification Plus status for commitment to diversity in leadership [News release].
Federal Reserve Board. (2024, July 10). Federal Reserve Board fines Citigroup $60.6 million for violating the Board's 2020 enforcement action [Press release].
Freshfields. (2026, February 27). FTSE Women Leaders Review 2026: Achieving gender balance five years on [Blog post].
Glass, C., & Cook, A. (2016). Leading at the top: Understanding women's challenges above the glass ceiling. The Leadership Quarterly, 27(1), 51–63.
Gupta, V. K., Han, S., Mortal, S. C., Silveri, S., & Turban, D. B. (2018). Do women CEOs face greater threat of shareholder activism compared to male CEOs? A role congruity perspective. Journal of Applied Psychology, 103(2), 228–236.
Gupta, V. K., Mortal, S. C., Silveri, S., Sun, M., & Turban, D. B. (2020). You're fired! Gender disparities in CEO dismissal. Journal of Management, 46(4), 560–582.
Haslam, S. A., & Ryan, M. K. (2008). The road to the glass cliff: Differences in the perceived suitability of men and women for leadership positions in succeeding and failing organizations. The Leadership Quarterly, 19(5), 530–546.
Hinchliffe, E. (2022, May 23). The number of women running Fortune 500 companies reaches a record high. Fortune.
Hinchliffe, E. (2026, June 3). GM is no longer the highest-ranked female-led company on the Fortune 500. Fortune.
Hinchliffe, E., & Abrams, J. (2024, June 4). The share of Fortune 500 businesses run by women can't seem to budge beyond 10%. Fortune.
Hollerith, D. (2024, July 10). Regulators fine Citigroup $136 million in setback for CEO Jane Fraser. Yahoo Finance.
Kulich, C., Ryan, M. K., & Haslam, S. A. (2007). Where is the romance for women leaders? The effects of gender on leadership attributions and performance-based pay. Applied Psychology, 56(4), 582–601.
Kulich, C., Ryan, M. K., & Haslam, S. A. (2014). The political glass cliff: Understanding how seat selection contributes to the underperformance of ethnic minority candidates. Political Research Quarterly, 67(1), 84–95.
Li, X., Moyer, O., Pfeiffer, O., Velthuis, R., & Zaleski, P. (2025). Female CEO selection: Does the glass cliff exist? The Leadership Quarterly, 36(2), Article 101853.
Matsa, D. A., & Miller, A. R. (2013). A female style in corporate leadership? Evidence from quotas. American Economic Journal: Applied Economics, 5(3), 136–169.
McCann FitzGerald. (2025). The Gender Balance on Listed Company Boards Directive takes effect in Ireland [Briefing].
Mohan, P. (2015, July 6). Reddit users push for CEO Ellen Pao's resignation. Fast Company.
Mohan, P. (2026, October 2). What a McKinsey partner's commentary on women CEOs reveals. Fast Company.
Morgenroth, T., Kirby, T. A., Ryan, M. K., & Sudkämper, A. (2020). The who, when, and why of the glass cliff phenomenon: A meta-analysis of appointments to precarious leadership positions. Psychological Bulletin, 146(9), 797–829.
Paul Weiss. (2024, December). Fifth Circuit strikes down Nasdaq board diversity disclosure requirements [Client memorandum].
Pereira, S. (2026, October 13). McKinsey removes partner over glass cliff LinkedIn row. Gulf News.
Reinwald, M., Zaia, J., & Kunze, F. (2023). Shine bright like a diamond: When signaling creates glass cliffs for female executives. Journal of Management, 49(3), 1005–1036.
Rink, F., Ryan, M. K., & Stoker, J. I. (2013). Social resources at a time of crisis: How gender stereotypes inform gendered leader evaluations. European Journal of Social Psychology, 43(5), 381–392.
Ryan, M. K., & Haslam, S. A. (2005). The glass cliff: Evidence that women are over-represented in precarious leadership positions. British Journal of Management, 16(2), 81–90.
Ryan, M. K., & Haslam, S. A. (2009). Glass cliffs are not so easily scaled: On the precariousness of female CEOs' positions. British Journal of Management, 20(1), 13–16.
Ryan, M. K., Haslam, S. A., Hersby, M. D., & Bongiorno, R. (2011). Think crisis–think female: The glass cliff and contextual variation in the think manager–think male stereotype. Journal of Applied Psychology, 96(3), 470–484.
Ryan, M. K., Haslam, S. A., & Kulich, C. (2010). Politics and the glass cliff: Evidence that women are preferentially selected to contest hard-to-win seats. Psychology of Women Quarterly, 34(1), 56–64.
Ryan, M. K., Haslam, S. A., Morgenroth, T., Rink, F., Stoker, J., & Peters, K. (2016). Getting on top of the glass cliff: Reviewing a decade of evidence, explanations, and impact. The Leadership Quarterly, 27(3), 446–455.
Silverman, L. (2017, June 13). Yahoo CEO Marissa Mayer resigns as Verizon completes $4.5 billion purchase. NPR.
The Globe and Mail. (2025, July 9). Linda Yaccarino resigns as X CEO. The Globe and Mail.
Utah State University. (2013). Studies show women and minority leaders have shorter tenures, tenuous support [News release].
Vollmer, L. (2004, December 1). Anne Mulcahy: The keys to turnaround at Xerox. Stanford Graduate School of Business.
Watkins, M. D., & Duke, L. S. (2014). Honeymoon interrupted: Mary Barra, CEO, GM (A) (Case No. IMD-7-1565). IMD.
Jonathan H. Westover, PhD, Chief Research Officer (Nexus Institute for Work and AI); Co-Founder & Chief Workforce and Learning Officer (Future State University); Founder & CEO (Human Capital Innovations); Professor of Organizational Leadership & Change (UVU). Read Jonathan Westover's executive profile here.
Suggested Citation: Westover, J. H. (2026). Standing at the Edge: What the Glass Cliff Evidence Really Says and How Boards Can Set Crisis Leaders Up to Succeed. Human Capital Leadership Review, 40(2). doi.org/10.70175/hclreview.2020.40.2.7
