Most companies are unhappy with their performance management. Only about 2% of organisations believe their approach works well, according to Deloitte's Human Capital Trends. And a 2026 paper published in Human Resource Management goes further. Murphy, who has studied performance appraisal for over 40 years, argues that the entire premise of performance management is flawed: its goals are destructive, its assumptions unlikely to be true, and its core procedures do more harm than good.
His conclusion is blunt: there is surprisingly little empirical evidence that performance management works, or that it has any reliable effects on the performance of employees whose performance is allegedly being managed. He cites DeNisi and Murphy's own 2017 review in the Journal of Applied Psychology as well as Pulakos and O'Leary's 2011 analysis that reached the same conclusion.
The measurement problem runs deeper still. Back in 1997, March and Sutton published a landmark paper in Organization Science arguing that most studies of organisational performance cannot identify true causal relationships. The performance advantage is unstable because effective practices are imitated. The causal complexity surrounding performance is enormous. And the data researchers rely on, mostly retrospective accounts from informants, are shaped by cognitive biases and conventional story lines rather than objective reality.
So which tools for performance management actually have evidence to back them up? This article examines every major tool through the lens of peer reviewed research. It also considers the critiques from Murphy, March, and Sutton, which should make any honest practitioner cautious about big claims.
Goal Setting: The Strongest Tool, but Watch the Side Effects
Goal setting is the single most powerful performance intervention available to managers. Locke and Latham summarised 35 years of goal setting research in American Psychologist. They reported that specific, difficult goals produced effect sizes (d) ranging from 0.52 to 0.82 across multiple meta analyses. A d of 0.82 is among the largest effects in organisational research. Employees with clear, challenging goals significantly outperform those given the directive to do their best.
Locke and Latham identified four mechanisms: goals direct attention toward relevant activities, increase effort, increase persistence, and, on complex tasks they trigger the development of new strategies. These findings come from research involving nearly 40,000 participants.
Murphy's 2026 paper raises a problem the goal setting literature often glosses over. Performance management systems use cascading goals to align every individual's work with strategic decisions made at the top. Murphy argues this is fundamentally a system of top down control that reduces worker autonomy and responsibility
Murphy's practical observation is that strategic goals become less relevant as they cascade down to individual workers. In many jobs, the tasks themselves don't fundamentally change based on broad strategy. March and Sutton support this scepticism from a different angle: any organisational practice that provides a competitive advantage is ordinarily adopted by competitors through imitation. Once everyone cascades goals, the practice ceases to be a differentiator.
Kleingelds, van Mierlo, and Iken, in a meta analysis published in the Journal of Applied Psychology covering 49 effect sizes, found that specific difficult group goals produced a large effect on group performance (d = 0.80). But egocentric individual goals within interdependent teams actually harm group performance (d =-1.75). Goals aimed at maximising the individual's contribution to the team worked well (d = 1.20).
Related: HR KPIs: Metrics for Driving Organizational Success
Feedback: The Tool That Backfires More Often Than You Think
Feedback is the centrepiece of most performance management systems. Kluger and DeNisi, in a meta analysis published in Psychological Bulletin covering 607 effect sizes and 23,663 observations, found that feedback improved performance on average (d = 0.41). But over one third of feedback interventions actually decreased performance. Murphy's 2026 paper highlights that over 90% of the positive effect sizes in that meta analysis were small, at d of 0.20 or smaller. Large positive effects from feedback are quite rare.
Murphy takes direct aim at frequent informal feedback, the signature tool of modern performance management. Advocates argue annual feedback comes too late to help. But Murphy argues the logic of more is better ignores the well established costs. Cleveland, Lim, and Murphy documented that negative feedback leads to resentment, harms supervisor subordinate relationships, creates stress, and damages employees' self image. Even when feedback is fair, these costs are real.
Even positive feedback creates problems. Mabe, P. A., & West, S. G. (1982). in the Journal of Applied Psychology established that people routinely overestimate their own performance. Because self assessments are inflated, even positive feedback is often received as disappointing. Murphy notes that feedback perceived as not positive enough leads to negative reactions and sometimes higher turnover, citing Kwak and Choi's 2015 research on rating discrepancy and turnover.
Murphy raises a question rarely addressed: who actually benefits from feedback? If employees know their goals and understand the metrics, they should already know whether they're meeting standards. Telling employees things they already know is unlikely to help.
March and Sutton add another dimension. They described how performance feeds back on itself through self reinforcing cycles. Good performance leads to positive attributions, greater confidence, and favourable treatment, which support future good performance. Poor performance triggers the opposite. Adding deliberate feedback on top of these powerful natural cycles may create interference rather than improvement.
Kluger and DeNisi proposed Feedback Intervention Theory to explain the pattern. Feedback that focuses attention on the task tends to improve performance. Feedback that shifts attention to the self tends to reduce it. Telling someone "your report lacked structure, here is how to fix it" works. Telling someone "you're underperforming" is a coin flip.
Braddy, Sturm, Atwater, Smither, and Fleenor's meta analysis of feedback orientation across 46 independent samples (12,478 workers), published in Human Resource Management Review, found feedback orientation correlates with work performance at r = 0.35. That's a meaningful relationship. But it also means some people are wired to use feedback well while others aren't.
Neubert, in a meta analysis published in Human Performance, found that adding feedback to goal setting produced d = 0.63 over goal setting alone. On complex tasks the effect more than doubled. Feedback works best not as a standalone tool but as a companion to clear goals.
360 Degree Feedback: Small Gains Unless You Add Coaching
Smither, London, and Reilly, in a meta analysis published in Personnel Psychology covering 24 longitudinal studies, found that improvement following multisource feedback was generally small, around d = 0.15. Murphy cites this work in his broader argument that feedback is a compromised tool.
The story improves with coaching. Smither, London, Flautt, Vargas, and Kucine, also in Personnel Psychology, found leaders who worked with a coach after receiving 360 feedback showed greater improvement than those who received feedback alone. Personality also matters: leaders high in emotional stability and conscientiousness were more likely to act on feedback.
Performance Appraisals: 30 Years of Measuring the Wrong Things
Schleicher, Baumann, Sullivan, and Yim, in a 30 year review published in the Journal of Applied Psychology in 2019, reviewed 488 studies spanning 1984 to 2018. Most focused on rating accuracy and employee satisfaction. Very few measured whether performance management changes behaviour, improves learning, or lifts organisational results.
March and Sutton anticipated this problem. They argued that performance data relies heavily on retrospective recall by informants who reconstruct the past to fit conventional storylines. Studies that claim to show what caused performance differences may tell us more about memory and storytelling than about the actual drivers.
Murphy's 2026 paper adds that even traditional appraisal, with all its faults, has advantages over modern performance management. Appraisal provides formal evaluation on multiple dimensions, an opportunity for employee input, and a link between performance and rewards. Informal check in systems often leave employees unclear about how performance connects to pay decisions.
Schleicher, Baumann, Sullivan, Levy, Hargrove, and Barros Rivera's 2018 systems review in the Journal of Management made a related point: performance management fails when components are treated in isolation rather than understood as a system.
Related: Mastering Performance Appraisal Software: Unlock Organizational Excellence
OKRs and Continuous Check ins: Popular but Understudied
Objectives and Key Results build directly on Locke and Latham's goal setting theory. In theory they should work because the underlying science is strong. In practice, Ordonez, Schweitzer, Galinsky, and Bazerman argued in the Academy of Management Perspectives that the side effects of goal setting, including narrowed focus, unethical behaviour, and damaged cooperation, were being routinely ignored.
Murphy adds that if cascading goals provide little meaningful guidance to individual workers, OKRs built on cascaded objectives inherit the same weakness. Continuous check ins lack the rigorous longitudinal evidence that goal setting and feedback have accumulated. Organisations adopting these methods should understand they are applying well supported principles through a format that hasn't been independently validated.
Coaching: What Murphy Says Organisations Should Do Instead
Murphy doesn't just criticise performance management. Drawing on the Ohio State leadership studies (Judge, Piccolo, and Ilies, 2004 in the Journal of Applied Psychology), he argues organisations should shift from managing performance to supporting it. Leaders should focus on providing consideration and support rather than constant monitoring and feedback.
The coaching evidence supports this. Theeboom, Beersma, and van Vianen, in a meta-analysis published in the Journal of Positive Psychology examining 18 studies, found that coaching produced significant effects on performance (d = 0.60), wellbeing (d = 0.46), and goal-directed self-regulation (d = 0.74).
Jones, Woods, and Guillaume, in a meta analysis published in PLOS ONE covering 17 studies, found coaching improved goal attainment (d = 0.74), self efficacy (d = 0.40), and resilience (d = 0.57). Effects were stronger when coaching lasted longer and used evidence based methods.
Murphy also cites Cannon Bowers, Bowers, Carlson, Doherty, Evans, and Hall's 2023 meta analysis in Frontiers in Psychology, which found coaching had a moderate positive effect on work outcomes. The direction of evidence is clear: supporting employees produces better results than trying to control their performance through feedback.
What the Evidence Actually Shows
Here is what the research tells us; Goal setting produces d = 0.52 to 0.82 (Locke and Latham, 2002). Feedback improves performance on average by d = 0.41 but decreases it in over a third of cases, with over 90% of positive effects at d = 0.20 or smaller (Kluger and DeNisi, 1996). Feedback combined with goal setting adds d = 0.63 over goal setting alone (Neubert, 1998).
360-degree feedback alone yields roughly d = 0.15 (Smither, London, and Reilly, 2005). Coaching produces d = 0.60 on performance (Theeboom and colleagues, 2014) and d = 0.74 on goal attainment (Jones and colleagues, 2016
And the overall effect of performance management in public sector organisations? Gerrish's 2016 meta-analysis of 49 studies in the Public Administration Review found the average effect was very weak. Murphy cites this in his 2026 paper to argue that, taken as a system, performance management has not demonstrated that it works.
March and Sutton remind us to be cautious even with these numbers. Performance advantage is competitively unstable. Effective practices get imitated. Retrospective data is biased. The honest conclusion is that some tools work better than others, but none work as reliably as their advocates suggest.
How to Build a System That Matches the Evidence
Start with goal setting, but use it carefully. The evidence from Locke and Latham for specific, challenging goals is unambiguous. But avoid rigid cascading that strips workers of autonomy. Murphy is right that goals imposed from above as a control mechanism create resistance, not engagement. Set goals collaboratively. Make them groupcentric in team settings.
Use feedback sparingly and focus it on the task. The Kluger and DeNisi finding should be on every manager's wall: feedback that directs attention to the self makes things worse. And remember Murphy's point: if performance goals and metrics are clear, employees often don't need to be told what they already know.
Replace monitoring with coaching. This is Murphy's central recommendation and the meta analytic evidence supports it. Coaching at d = 0.60 (Theeboom and colleagues) outperforms feedback at d = 0.41 with a third of effects negative (Kluger and DeNisi) and far outperforms 360 degree feedback alone at d = 0.15 (Smither and colleagues).
Measure what matters, but be honest about what measurement can tell you. March and Sutton showed that the causal stories we tell about performance are usually simpler and more confident than the evidence warrants. Track behaviour change and learning, not just satisfaction scores.
Related: Job Characteristics Model: What 50 Years of Research Actually Found
Support Performance. Don't Try to Manage It.
Murphy's 2026 paper ends bluntly: it is time to abandon the illusion of performance management. March and Sutton, writing nearly 30 years earlier, reached a related conclusion about performance research itself. These two papers, separated by three decades, converge on the same uncomfortable truth: our confidence in understanding what drives performance has always exceeded our actual knowledge.
That doesn't mean giving up. The tools with the strongest evidence, goal setting, task focused feedback, and coaching, all work best when they support employee autonomy. The weakest tools are the ones that monitor and enforce compliance.
The next time you evaluate tools for performance management, ask this: does this tool help my managers support better performance, or does it help them police compliance? The research is clear about which approach works.
