In the 1980s, Jack Welch institutionalized a practice at General Electric that would influence performance management for decades: every year, managers were required to identify their top 20% of performers (who were rewarded generously), their middle 70% (who were maintained), and their bottom 10% (who were let go). The practice became known as “rank and yank” or, more formally, stack ranking employees — a forced distribution system that ranks all employees against each other and takes differential action based on where they land. It spread widely. It also produced some of the most documented failures in the history of corporate talent management, culminating in Microsoft’s public abandonment of the practice in 2013 after it was blamed for years of internal dysfunction and competitive stagnation.
What Is Stack Ranking?
Stack ranking is a performance management approach that requires managers to rank all employees in their team or organization from highest to lowest performer, or to distribute ratings according to a forced curve (e.g., 20% top performers, 70% middle, 10% low performers). Unlike rating systems that evaluate each employee against a defined standard, stack ranking evaluates employees against each other. The distribution is predetermined: regardless of how strong the actual team is, someone must always fall in the bottom category. Stack ranking is also called forced distribution, forced ranking, or vitality curves.
Why Stack Ranking Was Adopted
Stack ranking addressed a real problem: rating inflation. In organizations without forced distributions, managers consistently rate most employees above average. This happens for entirely understandable psychological reasons — managers want to maintain positive relationships, avoid difficult conversations, and avoid creating administrative work around underperformance. The result is a rating system where 80–90% of employees are rated “above average” or higher, which tells no one anything useful.
Stack ranking’s appeal was that it forced differentiation. If you know that only 20% of your team can receive a top rating, you have to be much more precise about who those 20% actually are. At GE, this discipline was credited with building a culture of high performance and with surfacing underperformers who might otherwise have coasted indefinitely. For a while, it worked — at least in large organizations with enough scale that a forced distribution could be statistically reasonable.
Why Stack Ranking Employees Fails: The Evidence
It Destroys Collaboration
The most thoroughly documented failure mode of stack ranking is its effect on collaboration. When employees know that their rating is determined relative to their peers rather than against an absolute standard, cooperation becomes irrational. Helping a colleague improve their work directly reduces your own competitive position. The most famous account of this dynamic comes from former Microsoft employees, who told reporters that their strategy shifted from competing against other technology companies to competing against their colleagues. Team members stopped sharing information, withheld expertise, and in some documented cases actively undermined peers’ work to protect their own relative standing. Kurt Eichenwald’s 2012 Vanity Fair piece titled “Microsoft’s Lost Decade” drew a direct line between stack ranking and the company’s failure to capitalize on early advantages in smartphones, tablets, and search.
It Damages High-Performing Teams Most
Stack ranking’s arithmetic is brutal in a counterintuitive way: it is most damaging to the best teams. A team where every member is genuinely high-performing must still produce a bottom 10%. When the weakest member is performing at a level most organizations would consider strong, the forced distribution creates an arbitrary loser. Over time, high performers learn to avoid being assigned to high-performing teams because the competition is stiffer. The predictable result is that excellent teams become less attractive to join — exactly the reverse of the incentive a performance management system should create.
It Produces Short-Term Thinking
Stack ranking creates pressure to produce visible individual results quickly — because relative rankings are typically determined over short cycles. This systematically disadvantages long-horizon work: research, infrastructure investment, team development, and mentoring. The employee who spends six months mentoring a junior colleague is investing in the team’s long-term output. In a stack ranking system, that investment typically shows up as reduced individual output in the short term, which hurts their relative ranking. Stack ranking selects against the behaviors most critical to organizational learning and long-term capability building.
It Creates Legal and Diversity Risks
Forced ranking systems have been the subject of numerous employment discrimination lawsuits in the United States and other jurisdictions. When forced distributions systematically produce lower ratings for protected classes — women, minorities, older workers — the organization faces significant legal exposure, even if no discriminatory intent exists. Ford Motor Company paid $10.5 million to settle age and race discrimination lawsuits related to its forced ranking system in 2001. The statistical mechanism of forced distributions amplifies any existing bias in the rating process: if unconscious bias causes managers to rate women slightly lower on average, a forced distribution requiring a fixed percentage in the bottom tier will concentrate women there disproportionately.
Where Stack Ranking Still Makes Some Sense
A blanket condemnation of any form of performance differentiation misses an important nuance. The problem with stack ranking as traditionally implemented is not that it differentiates — it is that it forces differentiation relative to peers rather than against standards, and it requires a fixed distribution regardless of actual team quality. Several organizations have found middle-ground approaches that preserve the useful pressure for differentiation while avoiding the worst failure modes.
Calibration without forced curves: Organizations can require managers to explicitly justify any rating above a threshold without imposing a predetermined distribution. This creates upward pressure against inflation without forcing a fixed bottom tier. Managers must explain their top ratings rigorously, but no one is automatically disadvantaged by being on a strong team.
Investment tiering based on observed behavior: Rather than ranking employees against each other, some organizations tier development investment (stretch assignments, development budgets, accelerated promotion tracks) based on observed behaviors. This preserves differentiation in rewards and investment without creating explicit rankings that pit employees against each other.
What High-Performing Organizations Use Instead of Stack Ranking
Continuous Performance Conversations
The alternative that has gained the most traction since the decline of stack ranking is a shift from periodic comparative ranking to continuous performance conversations. Companies like Adobe — which eliminated annual performance reviews entirely in 2012 and replaced them with frequent “check-ins” — report significant improvements in voluntary turnover (down 30%) and manager satisfaction. The continuous model addresses the inflation problem not by forcing comparisons, but by building a culture where feedback is normal, frequent, and developmental rather than evaluative and comparative.
Calibrated Rating Standards With Behavioral Anchors
Behavioral anchors — explicit descriptions of what each rating level looks like in practice — address rating inflation by making it harder to justify an inflated rating without specific behavioral evidence. A manager who wants to rate an employee “Exceptional” must describe specific behaviors that match the exceptional anchor. This creates accountability for ratings without requiring relative comparison. Pair this with calibration sessions where multiple managers review ratings together, and you get the differentiation pressure of stack ranking without the relative ranking dynamic. See our guide to performance calibration meetings for how to run these effectively.
The 9-Box Grid for Development, Not Ranking
The 9-box grid plots employees on performance versus potential, but unlike stack ranking, it does not require a fixed distribution and does not tie directly to termination or retention decisions. Used well, it guides development investment: employees in different boxes receive different types of development support, not different termination risk. It preserves the value of distinguishing between performance levels while avoiding the existential competition that stack ranking creates.
OKR-Based Goal Achievement
Evaluating employees primarily against their own OKR achievement rather than relative to peers creates differentiation based on standard attainment rather than comparative ranking. An employee who achieves 100% of ambitious, well-scoped OKRs is clearly performing at a high level regardless of how their teammates performed. This approach is only valid if goal-setting quality is high — if goals vary widely in difficulty and ambition across employees, OKR achievement becomes a measure of goal-writing skill rather than actual performance.
Frequently Asked Questions About Stack Ranking
Do any major companies still use stack ranking?
Is there a difference between stack ranking and a bell curve rating distribution?
How do you prevent rating inflation without stack ranking?
Bottom Line
Stack ranking employees was designed to solve a real problem — rating inflation and insufficient performance differentiation — but its method of forcing relative comparison produced predictable and serious consequences: destroyed collaboration, disadvantaged high-performing teams, encouraged short-term thinking, and created legal exposure. Most high-performing organizations have moved to alternatives that preserve differentiation without comparative ranking: continuous feedback cultures, calibrated behavioral rating standards, 9-box talent planning, and OKR-based goal achievement. The goal is not to eliminate accountability but to ground it in standards rather than in competition with peers.
