Ask a manager to rate an employee's annual performance. Then ask them to list five specific achievements from Q1 without looking at any notes. In our internal research across 250+ performance review conversations, fewer than 18% of managers could do it unprompted. Most rated confidently anyway — their scores driven by what they remembered from the last 90 days.
This is the recency trap: the well-documented cognitive phenomenon where events that happened recently feel more important, more salient, and more representative than events that happened further back. In everyday life, this is mostly harmless. In performance management, it systematically disadvantages employees whose best work happened in Q1 or Q2, and systematically rewards those who were visible in Q4.
The Research Behind Recency Bias
The recency effect was first described by Hermann Ebbinghaus in 1885 as part of his forgetting curve research. In a business context, it was applied to performance management by researchers in the 1970s and 1980s, who found that supervisory ratings were disproportionately influenced by the most recent observation period — regardless of the overall performance trajectory.
More recent research has found that the effect is stronger under two conditions: when the review interval is longer (annual reviews produce more recency bias than quarterly check-ins), and when the reviewer lacks structured documentation (managers who kept no performance notes showed significantly stronger recency effects than those who maintained running records).
Research shows annual reviews produce recency bias 3–4× stronger than quarterly check-ins. The fix is not more frequent reviews alone — it is continuous, structured evidence capture throughout the year.
Who Gets Hurt Most
The recency trap does not hurt everyone equally. It disproportionately affects employees who deliver sustained, reliable performance throughout the year but are less visible in Q4. Field workers, remote employees, those in support functions, and employees who do not self-promote tend to see their annual contributions erased by a quieter final quarter.
By contrast, employees who are physically proximate to managers, who work on projects that peak in Q4, or who are skilled at ensuring their achievements are visible at year-end tend to systematically outperform their actual contribution in annual ratings — not because they worked harder, but because they were remembered better.
Breaking the Trap Structurally
- Continuous OKR tracking: When goal progress is recorded in real time, managers have an evidence trail to consult rather than relying on recall.
- Quarterly manager check-ins with documented outputs: Brief structured conversations throughout the year create multiple memory anchors, diluting the recency effect.
- AI-assisted review facilitation: Systems like TARA can prompt managers with documented evidence from early in the year before ratings are submitted.
- Calibration based on evidence, not impression: Peer calibration sessions that require managers to cite specific examples before finalising ratings.
The recency trap is not fixed by asking managers to 'be fairer.' It is fixed by building a process that makes the full year's evidence as accessible and salient as last month's. That is an infrastructure problem, not a training problem.