Case Study 2.2: The Performance Review — Seeing and Being Seen at Work

Overview

This case study examines the annual (or continuous) employee performance review as a panoptic institution — a site of visibility, individuation, normalization, and self-discipline that most workers inhabit for decades of their professional lives. It combines analysis of the traditional performance review with examination of newer continuous monitoring and real-time feedback systems that are replacing or supplementing annual reviews in many organizations.

Estimated Reading and Analysis Time: 60–75 minutes

Background: The Performance Review as Surveillance Ritual

The formal employee performance review is one of the most widespread surveillance rituals in contemporary life. In the United States alone, surveys suggest that approximately 70–80% of organizations conduct formal annual or semi-annual performance reviews. Globally, performance assessment processes — in various forms — exist in virtually every institutional context that employs people.

The basic structure is familiar: at intervals defined by the organization, a supervisor assesses an employee's performance against criteria that are typically determined by the organization, documents the assessment in writing, communicates it to the employee in a formal meeting, and files the documentation in a personnel record that may influence future decisions about compensation, advancement, assignment, and termination.

From the employee's perspective, the performance review is the moment when the gaze that has been watching them throughout the review period is formalized, documented, and made consequential. It is, in Foucauldian terms, the examination — the moment when surveillance produces a record that enters an individual's permanent file.

The Examination as Surveillance Technology

Foucault's analysis of the examination in Discipline and Punish is one of his most precise analyses of how surveillance mechanisms produce both knowledge and power:

"The examination combines the techniques of an observing hierarchy and those of a normalizing judgement. It is a normalizing gaze, a surveillance that makes it possible to qualify, to classify, and to punish... The examination is at the centre of the procedures that constitute the individual as effect and object of power, as effect and object of knowledge."

The performance review is the workplace examination. Let us trace Foucault's three effects through it:

Individuation: The performance review creates an individual file — a dossier of assessments, incidents, ratings, and written evaluations attached to the specific worker. Before the modern performance management system, workers might have been managed as groups (the entire production team is meeting its quota) or through direct supervision (the foreman watches and gives direct instruction). The performance review system makes each worker a distinct object of knowledge, classifiable relative to a population of peers.

Permanence: The documented performance review enters a personnel file that may persist for the employee's entire tenure — and, in some cases, longer (references, background checks, internal HR systems). An employee who had a difficult review ten years ago may find it referenced in a current investigation or promotion decision. The permanence effect means that the employee is always, at some level, performing for the historical record.

Visibility: The performance review makes the employee visible to levels of organizational hierarchy they do not normally interact with. In many systems, reviews are reviewed by the supervisor's supervisor, by HR, and by compensation committees. The employee is visible at these levels not as a whole person but as a data profile — a set of ratings, written comments, and comparative rankings.

Stack Ranking and Normalization

One of the most controversial implementations of performance review systems is "stack ranking" — the practice of forcing a distribution of performance ratings, requiring that a certain percentage of employees be rated in each category (e.g., "top performer," "meets expectations," "needs improvement"), regardless of actual performance variation within the group.

Microsoft's "stack ranking" system, which required that a fixed percentage of employees in each team be rated at the lowest performance tier, was widely reported as having produced significant organizational dysfunction — employees competing with each other rather than collaborating, incentives to join teams where competition was weaker, and a systematic disadvantage for any employee unlucky enough to be assigned to a high-performing team where they would face a more competitive ranking context.

Microsoft abolished its stack ranking system in 2013, to widespread internal relief.

The stack ranking system is a pure Foucauldian normalization mechanism: it defines the distribution of normal performance by administrative fiat and then classifies each individual relative to that distribution. The classification drives consequences. The distribution is real — people lose jobs because of it — but the categories are constructed.

The Rise of Continuous Performance Management

Many organizations are moving away from the annual review toward "continuous performance management" — frequent check-ins, real-time feedback apps, and ongoing goal tracking that replaces the once-yearly examination with a permanent stream of evaluation data.

Tools like Lattice, Betterworks, and Culture Amp facilitate continuous feedback loops: managers rate employee performance on short timescales, peers provide real-time feedback, and employees themselves log their progress toward goals in systems visible to their managers.

The stated benefit is that continuous feedback is more timely and actionable than annual reviews. The panoptic implication is that the visibility regime is no longer concentrated in two days per year (pre-review and review day) but is ongoing. The employee is always being observed, always generating data, always potentially subject to assessment.

The Peer Review Problem

Many continuous performance management systems incorporate peer feedback — ratings and written comments from co-workers. The stated purpose is to give employees a fuller picture of their performance from multiple perspectives.

The surveillance implication is that the peer review transforms horizontal relationships (co-workers) into vertical surveillance relationships (co-workers who assess and are assessed by each other). This is precisely the distributed surveillance effect that Foucault analyzed as characteristic of disciplinary power: the inspector is no longer a single figure in a tower but is distributed across the social body, each member of which survives partly by monitoring the others.

Algorithmic Performance Management: The Next Phase

In logistics warehouses, call centers, and delivery services, algorithmic performance management has partly or wholly replaced the human supervisor's review. As discussed in Case Study 1.1, systems track worker metrics continuously and generate automated assessments that drive staffing, scheduling, and termination decisions.

This development extends the performance review's panoptic logic to its extreme: instead of a once-yearly examination that produces a fixed record, there is a continuous examination that generates an always-updating score. The employee is never outside the performance review; the review is the permanent condition of employment.

The algorithmic dimension adds a new feature: opacity. The human performance review, for all its subjectivity, is conducted by a specific person whose reasoning the employee can (in principle) contest. The algorithmic review is conducted by a system whose reasoning the employee typically cannot access, contest, or understand. The panoptic visibility is asymmetric not just in that the watcher can see the watched, but in that the watched cannot see the logic by which they are being evaluated.

Racial and Gender Dimensions

Research on performance reviews has documented consistent patterns of bias:

Racial bias: Studies comparing performance reviews for equally performing employees from different racial groups have found that Black and Latino employees receive lower ratings and less specific developmental feedback than white employees with comparable performance. The structured rating system of performance reviews does not eliminate bias; it may systematize it.

Gender bias: Research on performance review language has found that women receive reviews more likely to focus on personality traits (emotional, likable, abrasive) while men receive reviews more likely to focus on work outcomes (accomplished, produced, led). Studies by companies including Google have found that women are less likely to receive "stretch assignments" reflected in performance goals, limiting career development.

The documentation asymmetry: Anecdotal and research evidence suggests that employees who are targets of discriminatory management are often more thoroughly documented than others — every minor infraction recorded, every behavioral issue noted — creating a paper trail that makes termination or demotion appear justified even when the underlying motivation is discriminatory. The performance review apparatus can be used to build a discriminatory case while appearing to operate as a neutral assessment system.

Connection to Chapter Themes

The performance review illustrates all three of Foucault's panoptic effects operating simultaneously in a single institutional mechanism. It also illustrates the limits of the panopticon metaphor:

Where the panopticon metaphor is apt: The visibility asymmetry (manager can see employee; employee cannot see manager's reasoning), the individuation (personal file, named and filed), the permanence (records follow the employee), and the normalizing function (the rating system defines what "good" performance looks like) are all classic panoptic elements.

Where the metaphor is strained: The peer review and continuous feedback systems suggest a more distributed, liquid surveillance mechanism — not a single inspector in a tower but a web of mutual assessment. The algorithmic management system involves a type of observation — data-driven, at scale, operating through inference — that the panopticon's architectural metaphor cannot fully capture.

Both the strengths and limits of the metaphor are instructive: they tell us what is continuous with earlier surveillance forms and what is genuinely new.

Discussion Questions

1. The Examination: Foucault says the examination "transforms each individual into a case." What does it mean for a worker to become a "case"? What is gained and lost in this transformation?

2. Purpose and Effect: Employers typically justify performance reviews as tools for employee development — helping workers understand their strengths, address weaknesses, and grow professionally. In your assessment, do performance reviews actually serve this purpose? What features of the system support or undermine this stated purpose?

3. Continuous vs. Annual: The shift from annual reviews to continuous performance management is presented by advocates as more employee-friendly — more timely feedback, more opportunities for course correction. Analyze this claim from a panoptic perspective. Is continuous feedback better or worse for employees, considered as subjects of a surveillance apparatus?

4. Algorithmic Opacity: The algorithmic performance management systems used in warehouses and delivery services make the evaluation logic opaque. Is this a meaningful difference from the human supervisor's assessment, which is also imperfectly transparent? What specifically is lost when the evaluating agent is an algorithm rather than a person?

5. Documenting Discrimination: The case study mentions that discriminatory management may be supported by selective performance documentation. How does visibility asymmetry — managers can see and document workers, but workers cannot see the decision logic — enable this dynamic? What structural changes would address it?

6. Self-Assessment: Many performance review systems require employees to complete a self-assessment before the formal review. Analyze the self-assessment component as a form of self-surveillance. Who benefits from the worker's self-disclosure? Is the self-assessment genuinely for the employee's benefit, or does it primarily generate data for management?

7. Jordan at Work: Jordan's warehouse job tracks their performance algorithmically. Imagine Jordan gets a desk job after graduation where performance is assessed through an annual review with a human manager. Which system is more panoptic? Which is more threatening to Jordan's interests? Are these the same question?

Chapter 2 | Case Study 2.2 | Part 1: Foundations | The Architecture of Surveillance