Every company says it wants feedback to flow upward, and almost every company has designed a system in which giving it honestly is a personal risk with no payoff. The annual engagement survey arrives, its questions carefully abstracted ("I have confidence in senior leadership"), its results aggregated into heat maps that no manager ever lost a job over. Employees answer diplomatically, managers skim defensively, and the ritual concludes until next year. Everyone involved knows it changes nothing, which is precisely why it is safe.
The research on employee silence explains the equilibrium. A 2022 meta-analysis by Hao and colleagues, aggregating 168 samples and more than 63,000 employees, found silence associated with worse job attitudes, lower well-being and worse performance, and found that silence is driven above all by inhibition: people scan for danger, and where psychological safety is missing, they conclude, accurately, that candor is a tax paid by the candid. Amy Edmondson's foundational field study of 51 work teams showed the same structure from the other side: teams learn when members believe interpersonal risk-taking is safe, and learning is the channel through which safety becomes performance. An upward-feedback form does not create that belief. Consequences do.
The Semco design
Semco's approach, as Ricardo Semler has told it, took the polite ritual and gave it teeth. Everyone rated their boss, anonymously, twice a year. A manager who scored badly twice in a row lost the role. The rule applied to Semler himself. And, crucially, the exit was engineered to be survivable: going back to technical work, at the same salary, was a legitimate outcome, not a walk of shame.
Each element answers a specific failure mode of conventional upward evaluation. Anonymity answers the silence problem: no one has to bet their career on a manager's graciousness. The twice-yearly cadence answers the memory problem: a single annual snapshot is noise, but two consecutive verdicts a few months apart are a signal. The consequence answers the cynicism problem: people report honestly when reporting does something. And the same-salary return answers the entrenchment problem, which is the one everyone else misses. If losing the role means losing income and face, every incentive pushes the failing manager to cling, politick and intimidate his way to better scores. Make the exit dignified and the grip loosens.
Underneath the mechanics sits the principle, which Semler states simply: someone who controls other people's days has to keep earning that control. Authority over humans is the one job input that is borrowed, not owned, and the people it is borrowed from are the only ones positioned to say whether it is being used well. The boss's boss sees the manager a few hours a month, mostly in performance mode. The team sees the weather daily.
What the ritual measures elsewhere
Run against this template, the standard corporate instrument reveals what it actually is: a sensor wired to no actuator. It collects the signal (usually blurred by aggregation into "leadership effectiveness indices") and routes it somewhere consequences cannot reach. Worse, in low-safety environments it functions as a periodic reminder that the company asks questions whose honest answers it does not want, which teaches silence more efficiently than never asking would.
The meta-analytic evidence suggests the cost of that lesson compounds. Silence is not the absence of noise; it is the absence of correction. Every quarter a struggling manager goes unrated, or rated without effect, the team's problems stay private, the manager's habits harden, and the organization keeps paying a person to make its people less effective, without ever receiving the invoice in legible form.
The uncomfortable minimum
Organizations do not need to adopt Semco's rule to learn from its structure, but they do need its three commitments in some form: protection for the rater, a consequence that a bad pattern reliably triggers, and an exit that lets the demoted keep their dignity and their salary. Remove any one and the system regresses to theater, because employees price consequences, not questionnaires.
The real test of upward evaluation is brutally simple: name the last manager in your company who lost the role because of what the team said. If no name comes to mind, the grades were never real. Only the silence was.
Sources
- Hao, Zhu, He, Duan, Zhao & Meng (2022). When Is Silence Golden? A Meta-analysis on Antecedents and Outcomes of Employee Silence. Journal of Business and Psychology. 168 samples, 63,134 employees: silence tracks worse attitudes, well-being and performance, driven by inhibition and lack of safety. https://link.springer.com/article/10.1007/s10869-021-09788-7
- Edmondson (1999). Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly. Field study of 51 teams: psychological safety enables learning behavior, which mediates team performance. https://journals.sagepub.com/doi/10.2307/2666999
- Semler (1989). Managing Without Managers. Harvard Business Review. First-person account of Semco's workplace democracy: profit sharing, open books, and workers' voice in management. https://hbr.org/1989/09/managing-without-managers