Exploration and Exploitation in Organizational Learning

James G. March · Original publication / lecture: 1991

Discusses the tension between discovering possibilities and exploiting existing capabilities.

Study the whole work ↓

Study the whole work

Follow the work’s structure and connect its central question, key concepts and assignments. Each unit’s reading questions are original tasks to consider with the work open. The edition and scope section identifies the materials checked and their limits.

The question of the whole work

How can models help us examine conditions under which learning improves current performance while narrowing future possibilities?

Edition and complete scope

Edition
Organization Science, volume 2, issue 1, 1991, pp.71–87. DOI:10.1287/orsc.2.1.71.
Scope
Opening, numbered sections, and references: 6 · Opening problem — References
Edition and editorial divisions
Six divisions cover the opening, sections 1–4, and references. Since section boundaries share pages, divisions follow verified sections rather than page ranges. Acknowledgments are supporting material at the end of section 4. Editorial parts and units do not convert conditional model results into universal organizational recommendations.
Checked
2026-11-17

Verification scope

Structure checked against: Public text

Checked INFORMS metadata and abstract against the original paper publicly hosted by San José State University, including four sections, two models, figure conditions, conclusion, and reference locations. Simulations were not reproduced and practical effects were not tested. Questions and diagramming tasks are editorial.

  1. Opening problem ↗
  2. 1 · The trade-off ↗
  3. 2 · Mutual learning ↗
  4. 3 · Competition ↗
  5. 4 · Model implications ↗
  6. References ↗

Choose a reading path

Follow the structure of the work

Open each part for its scope, account, key passages and questions. Close reading revisits passages; reading follows their development; a survey identifies their role in the whole.

Define the allocation problem

Opening problem — 1 · The trade-off

Read the opening and first section to distinguish exploration and exploitation by where and when costs and benefits occur.

Ask about two forms of learning

Opening problem — Opening problem · Read

Read the opening problem without classifying exploration and exploitation solely by activity names. Specify the resource being allocated and the outcomes being compared.

Passages to examine in the work

  • Opening problem ↗

    Reading question: Return to the definitions and consider when the same activity could serve a different role.

Question

Read the opening, identify activities that may involve exploration or exploitation, and retain one ambiguous case.

Viewpoints for an answer

  • Distinguish novelty of an activity from its role in developing existing competence.
  • Specify whose time or other costs are being allocated.

Key concepts:ExplorationExploitation

Compare distributed costs and benefits

1 · The trade-off — 1 · The trade-off · Close reading

Read section 1 by distinguishing timing, recipients within the organization, and external competition. Avoid immediately converting allocation into one fixed ratio.

Passages to examine in the work

  • 1 · The trade-off ↗

    Reading question: At the discussion of time and participants, inspect the assumptions in your own example.

Question

Choose two allocation difficulties from section 1 and construct an example evaluated differently in the short and long run.

Viewpoints for an answer

  • Show whether the people bearing costs are also those receiving the benefits.
  • Identify which assessment changes when the evaluation time changes.

Key concepts:ExplorationExploitationDistributed costs and returns

Trace learning by people and organization

2 · Mutual learning — 2 · Mutual learning

Read the assumptions, update rules, and results in section 2. Keep individual knowledge distinct from knowledge represented in the organizational code.

Builds on:Define the allocation problem

Read the mutual-learning model

2 · Mutual learning — 2 · Mutual learning · Close reading

In section 2, trace how the organizational code and individual beliefs change. Read learning rates, personnel turnover, and environmental change as distinct conditions.

Passages to examine in the work

  • 2 · Mutual learning ↗

    Reading question: Move between the model definition and figures, recording what condition changed and what was measured.

Question

Diagram who learns from whom in section 2, then explain one assumption that could change the resulting pattern.

Viewpoints for an answer

  • Track knowledge in the organizational code separately from knowledge of an average individual.
  • Distinguish newcomers’ amount of knowledge from its difference from existing knowledge.

Key concepts:Organizational codeSocializationModel assumption

Examine competition and model limits

3 · Competition — References

Read the competitive setting in section 3, then return to section 4 and the references. Distinguish average performance from relative rank and assumptions from practical decisions.

Builds on:Trace learning by people and organization

Distinguish performance from rank

3 · Competition — 3 · Competition · Close reading

Read the competitive model in section 3 by distinguishing higher mean performance from the chance of ranking near the top. Attend to distributions and the number of rivals.

Passages to examine in the work

  • 3 · Competition ↗

    Reading question: Read the competitive conditions attached to a figure and check which objective is affected by increased variability.

Question

Read section 3 and identify what must be compared under a mean-performance objective and a top-ranking objective.

Viewpoints for an answer

  • Present the mean, variability, and number of competitors as separate conditions.
  • Retain the distinction between seeking a high rank and avoiding a low rank.

Key concepts:Competitive ecologyModel assumption

Ask what small models establish

4 · Model implications — 4 · Model implications · Read

Read section 4 to compare what the two models share and where they differ. Separate questions suggested by a model from decisions requiring additional evidence.

Passages to examine in the work

  • 4 · Model implications ↗

    Reading question: Look for qualifications in the conclusion and check whether your proposal goes beyond them.

Question

Open section 4 and draft one workplace proposal, adding the local information required and what the paper alone cannot decide.

Viewpoints for an answer

  • Write the model result and your proposed action in separate sentences.
  • Address both reasons to sustain exploration and the costs exploration entails.

Key concepts:Model assumptionDistributed costs and returns

Investigate the research connections

References — References · Survey

Use the references to investigate the research areas connected to the paper. Distinguish an account of one paper from claims about agreement across a field.

Passages to examine in the work

  • References ↗

    Reading question: Move between the citation in the text and the reference entry to examine its role in the argument.

Question

Choose a reference, locate its use in the paper, and formulate one question for further investigation.

Viewpoints for an answer

  • Distinguish a work being cited from having checked its conclusions.
  • Clarify whether the next reading is for checking assumptions, comparison, or seeking counterexamples.

Key concepts:Model assumption

Return key concepts to the text

Exploration
Activities that investigate possibilities not yet well understood; compare their timing and uncertainty with exploitation. First unit to examine →
Exploitation
Activities that develop and use existing knowledge or competence; examine their function rather than labels alone. First unit to examine →
Distributed costs and returns
The occurrence of learning costs and benefits at different times and for different people or organizations. First unit to examine →
Organizational code
A representation of organizational beliefs in the mutual-learning model, not an automatic equivalent of a real manual or all collective knowledge. First unit to examine →
Socialization
In the model, the process by which individuals learn from the organizational code; examine its speed and effects on organizational learning separately. First unit to examine →
Model assumption
A condition imposed to make a complex phenomenon analyzable, whose relevance must be checked before applying results in practice. First unit to examine →
Competitive ecology
A setting where the value of performance depends on rivals’ performance and number; distinguish absolute outcomes from relative position. First unit to examine →

Avoiding misreadings

These are editorial examples of misreadings, with passages that help reconsider them.

  • Every organization should allocate equally to exploration and exploitation.

    Specify the time horizon and whose costs and returns are counted before evaluating any proposed ratio.

    1 · The trade-off ↗
  • Because turnover can help, more turnover always improves organizational knowledge.

    Check learning rates together with turnover conditions in section 2, and retain the conditional nature of the comparison.

    2 · Mutual learning ↗
  • Greater variability helps in every kind of competition.

    Check ranking objectives and numbers of competitors in section 3, distinguishing mean performance and avoiding low ranks.

    3 · Competition ↗

An assignment connecting the whole work

Choose a proposal for making short training more efficient and reconsider it through both models. Identify expected outcomes, differences that could be lost, conditions to check locally, and decisions the paper alone cannot settle.

  • Diagram who bears the costs and receives the benefits, and when.
  • Separate assessments of individuals, the organizational code, and competitive rank.
  • Connect each relevant model assumption with evidence to collect in practice.

Parts to connect:Define the allocation problemTrace learning by people and organizationExamine competition and model limits