Want Employees to Learn Fast Without Killing Exploration? Update the Scoreboard Faster.

Organizations face a familiar learning problem. When employees quickly copy today’s best performer, useful practices spread fast—but everyone can converge on the same approach before better alternatives are discovered. Classic organizational-learning models have therefore produced a counterintuitive prescription: slower learning can preserve diversity and improve long-run performance.

New research argues that the problem may not be learning speed itself. It may be stale performance information.

In “Fast learning and sustained exploration: The role of timely performance feedback,” a 2026 Strategic Management Journal article by Jerker Denrell, Michael Christensen, Thorbjørn Knudsen, and Chengwei Liu, a computational model shows that fast learning can support exploration when information about who is performing well is updated frequently.

The Research

The paper revisits a classic family of computational models inspired by James March’s work on organizational learning. Those models explore how individuals copy higher performers and how the speed of imitation affects diversity and long-run organizational performance.

Prior results suggested that fast learning creates premature convergence: everyone copies the same stars, practices become homogeneous, and the organization loses opportunities to discover better combinations.

Denrell and colleagues identify a critical assumption behind that result. In classic models, information about performance is not continuously refreshed. People can therefore imitate someone who was a top performer at the beginning of a period but is no longer superior by the time copying occurs.

When Feedback Is Current, Fast Learning Performs Better

The researchers modify the modeling environment so performance information can be updated more frequently.

Under those conditions, the advantage of slow learning largely disappears. Fast learners can adapt whom they copy as new performance information becomes available rather than continuing to follow stale stars.

The paper’s conclusion is not simply “copy faster.” It is that learning speed and feedback speed must be considered together. Rapid imitation based on old information can be dangerous; rapid learning based on timely information can improve long-run outcomes.

The Key Mechanism Is “Target Diversity”

Classic arguments emphasize preserving diversity in the overall population: keep people from copying one another too quickly so different practices survive.

The new model shifts attention to what the authors call “Target Diversity”—diversity in the people who become attractive imitation targets over time.

Fast learning allows initially weaker performers to copy useful components and move quickly toward the performance frontier. As they improve, some become top performers themselves. Their distinct knowledge combinations then become visible to others.

Instead of everyone repeatedly copying one established star, the organization gets a continually changing set of successful people to learn from.

Underdogs Can Become Valuable Sources of Knowledge

This mechanism produces an especially interesting implication. Employees who begin below the frontier may possess combinations of knowledge that established stars do not have.

If they can learn rapidly, catch up, and then become visible as soon as their performance improves, their unique approaches enter the organization’s pool of ideas worth imitating.

The organization can therefore exploit diversity rather than preserving it simply by limiting communication.

This is why the managerial summary recommends spotlighting emergent high performers regardless of their historical reputation.

Reputation Can Become a Learning Trap

Organizations often direct attention toward people or units with long track records of success. That seems sensible, but it can make the set of models employees learn from unusually stable.

The computational findings suggest a different design: use current performance information to identify who is succeeding now, including newly successful “underdogs.” Doing so can continually refresh the set of practices circulating through the organization.

The result is exploration generated through shifting learning targets rather than through deliberate isolation.

What Managers Should Do Differently

Know What the Evidence Is

The findings come from computational modeling, not a field experiment demonstrating that a specific dashboard or recognition program will produce the same effect in every company. The value of the study is in clarifying a mechanism and its design implications.

It shows that claims about the danger of rapid learning depend critically on how current performance information is and how imitation targets are selected.

The Takeaway

Organizations do not necessarily need to slow employees down to preserve exploration. Denrell, Christensen, Knudsen, and Liu’s model suggests a more managerially intuitive approach: let people learn quickly, but keep the performance information they use to choose whom to learn from up to date.

The goal is not to create one permanent best-practice star. It is to build a system in which new high performers can emerge, become visible, and contribute their distinctive knowledge to the next round of learning.

If everyone is copying yesterday’s winner, the solution may not be less learning. It may be a better scoreboard.