Forced Ranking, Calibration and Continuous Feedback: Choosing a Performance Model

Key takeaways
- The shift in rankings: Six percent of Fortune 500 companies have eliminated forced rankings, according to CEB (shrm.org).
- Executive skepticism: A public survey by Deloitte found that 58% of executives believe their current performance management approach drives neither engagement nor productivity (engagedly.com).
- Developmental vs. accountability models: Companies like Adobe and Microsoft lean toward development-first models, while Amazon and Meta lean toward accountability-first models using calibrated scores (engagedly.com).
- The time cost of reviews: An annual performance review can take three weeks of cumulative time per year, whereas feedback can be delivered in small segments regularly (shrm.org).
A significant portion of executive leadership is questioning the efficacy of traditional performance management. In a public survey, 58% of executives questioned by Deloitte believe their current performance management approach drives neither employee engagement nor productivity (engagedly.com). This sentiment aligns with a broader movement among large-scale employers to rethink how talent is measured and rewarded.
The decline of forced ranking
For decades, forced ranking was a standard tool used to identify top performers and manage out low performers. However, research indicates that forced rankings do not foster productivity or improvement and can create antagonism between managers and workers (shrm.org). When an organization mandates that a specific percentage of employees must fall into a low-performing category, the mechanism changes from measuring individual output to managing a fixed distribution.
What the data says: Six percent of Fortune 500 companies have eliminated rankings entirely, according to CEB (shrm.org).
What it does not say: The data from CEB does not specify which industries or specific organizational sizes are most likely to abandon these systems, nor does it track the long-term impact on retention rates following the removal of rankings.
The mechanism behind the antagonism is often rooted in the "zero-sum" nature of the model. If a manager has ten high-performing employees but is forced to rank two of them as "underperforming" to satisfy a quota, the incentive to collaborate diminishes. Employees may begin to view their colleagues as competitors for a limited number of high slots rather than as teammates working toward a common goal. This behavior often leads to information hoarding and a reduction in the very collaborative behaviors that drive modern organizational success.
Calibration versus continuous feedback
As organizations move away from rigid rankings, they generally gravitate toward one of two distinct philosophies: accountability-first or development-first. This distinction is visible in how major corporations structure their talent reviews.
| Company / Group | Primary Focus | Mechanism / Approach |
|---|---|---|
| Adobe, Microsoft, Deloitte | Development-first | No forced ratings (engagedly.com) |
| Amazon, Meta | Accountability-first | Calibrated scores (engagedly.com) |
| Accenture | Performance Achievement | Forward-looking conversations (engagedly.com) |
The accountability-first model, used by companies like Amazon and Meta, relies on calibrated scores to ensure rigor (engagedly.com). Calibration involves a group of managers discussing their direct reports to ensure that a "high performer" rating in one department means the same thing as a "high performer" rating in another. This process aims to reduce individual manager bias and ensure that compensation and promotions are tied to standardized metrics. The practical implication is that while the system is rigorous, it still requires significant managerial time to execute effectively.
The development-first model, utilized by Adobe, Microsoft, and Deloitte, prioritizes growth and removes forced ratings (engagedly.com). This approach assumes that the primary purpose of performance management is to improve the employee's capability rather than to sort them into buckets. The mechanism here is psychological; by removing the fear of a low rank, the organization seeks to encourage risk-taking and honest communication about skill gaps.
Managers can use real-time feedback to gain a broad, well-sourced perspective that can eliminate personal bias, according to Kris Duggan, CEO at BetterWorks (shrm.org).
Why It Matters
The mechanics of continuous feedback
The shift toward continuous feedback is often a response to the inefficiency of the annual cycle. Kris Duggan, CEO at BetterWorks, noted that an annual performance review takes three weeks a year, whereas it can be done in small segments regularly (shrm.org).
What the data says: Organizations moving away from annual reviews are often shifting toward providing ongoing feedback, according to Dominique Jones of Halogen Software (shrm.org).
What it does not say: The data does not show whether continuous feedback models lead to higher actual productivity or if they simply increase the frequency of administrative tasks for managers.
The mechanism of continuous feedback relies on the frequency of the interaction. When feedback is delivered once a year, it becomes a retrospective autopsy of past mistakes. When it is delivered weekly or monthly, it becomes a real-time course correction. David Brennan of Achievers suggests that employee performance should be judged against metrics set for success for a role rather than compared to others (shrm.org). This moves the focus from "how am I doing compared to John?" to "how am I doing against the requirements of my role?"
Analysis: Comparing the models
To choose a model, leaders must understand the trade-offs between rigor and agility. We can analyze these by looking at how they handle three critical business needs: compensation, bias reduction, and speed.
Analysis: Compensation and Data Reliability
One challenge with continuous feedback is the difficulty of translating a series of "check-ins" into a single, defensible number for a year-end bonus. Jason Averbook, CEO of TMBC, suggested that frequent questions to team leaders allow organizations to establish statistically reliable data for promotions, raises, and bonuses (shrm.org). This suggests that continuous feedback does not have to be purely qualitative; if the frequency is high enough, the data points can be aggregated to create a robust, evidence-based compensation decision.
Analysis: The Bias Trade-off
Calibration is a strong tool for reducing individual manager bias because it forces peer review of ratings. However, it can also create a "culture of comparison" if not managed carefully. Conversely, development-first models reduce the pressure of comparison but may struggle to identify "dead wood" or underperformers who require exit management. The choice often depends on whether the organization's primary risk is "lack of rigor" (not identifying low performers) or "lack of psychological safety" (employees being too afraid to fail).
Analysis: Administrative Load
The "three-week" annual review is a massive, concentrated burst of administrative work that often disrupts business operations during the review season. Continuous feedback spreads this load. However, the risk is "feedback fatigue," where managers feel they must constantly be "on" or documenting every interaction, which can lead to superficial or perfunctory feedback that lacks substance.
The role of technology in modern reviews
As companies move toward these models, technology is increasingly used to facilitate the process. For example, JPMorgan allows employees to use an internal AI tool to draft year-end reviews, though humans must own the final version (engagedly.com).
The mechanism here is augmentation, not replacement. The AI handles the heavy lifting of synthesizing a year's worth of written notes or data points, but the human manager remains the final arbiter of the performance assessment. This addresses the "time cost" problem mentioned earlier while maintaining the accountability that AI-only systems lack. However, organizations must be cautious; if the AI's training data contains historical biases, the drafted reviews may inadvertently replicate those biases, making the human "ownership" step critical for equity.
What HR can do on Monday morning
Transitioning from a legacy model to a modern performance framework is a multi-quarter project, but the foundation can be laid immediately. Instead of attempting a total overhaul of the entire company's philosophy overnight, HR teams can implement small-scale pilots to test the mechanisms described above.
If your organization is currently using forced rankings and is seeing signs of antagonism or low engagement, consider these immediate steps:
- Audit the time spent: Ask managers to track how many hours they spend on the annual review process versus how many hours they spend on one-on-one coaching. Use this to build the business case for a continuous feedback model.
- Define "Success Metrics" for roles: Before changing the review system, ensure every role has clear, role-specific metrics. As David Brennan suggests, performance should be measured against role success, not peer comparison (shrm.org).
- Pilot "Weekly Check-ins": Encourage managers to hold 15-minute weekly meetings with each employee. The goal is not a formal review, but a simple alignment on current priorities and blockers.
- Test the "Calibration" concept: Before implementing a formal calibration process, have managers in a single department review their ratings together to see where they disagree on what constitutes "exceeding expectations."
The goal for 2026 and beyond is to move from a system of "policing" performance to a system of "enabling" performance. Whether an organization chooses the accountability-led path of Amazon or the development-led path of Adobe, the effectiveness of the system will depend on whether the data collected is frequent, role-specific, and actionable.