Better decisions. More thoughtful work.

The Garbage Can Model Explains Why Some Decisions Feel Backwards

A team workshop with separate colourful cards for problems proposals and decisions.
AI-generated photographic illustration; not documentary photography.

A team approves a new tool, then spends the next meeting deciding which problem it will solve. The sequence feels backwards because it is. Yet it can happen without anyone deliberately trying to make a bad decision.

In their 1972 paper, Michael Cohen, James March and Johan Olsen proposed a garbage can model of organisational choice. It examines settings with unclear preferences, imperfectly understood processes and changing participation. Problems, available solutions, participants and opportunities to make choices do not always arrive in a neat order.

The label is a metaphor for that mixture, not an insult to employees. Nor is the model a recommendation to throw every issue into a meeting and hope for the best. Its value is in making a particular kind of disorder easier to notice.

Try it on a fictional proposal

Imagine a department has a software licence about to expire. One manager already wants a different product. Another brings a reporting problem. A third attends only the first half of the meeting. A renewal deadline creates the choice opportunity, and the preferred product becomes attached to the reporting problem before anybody has tested the fit.

That example is invented, but it supplies four useful questions:

  • Which problem was explicitly defined?
  • Which solution was already waiting for a use?
  • Who participated when the commitment was made?
  • What deadline or event created the opportunity to decide?

The answers do not prove the decision was wrong. They show where a follow-up check is needed. The tool may still be suitable, but the team should test that suitability rather than treating the meeting’s sequence as evidence.

A practical improvement is to record the problem and success criteria before evaluating the proposed solution. If a new problem appears halfway through, name the change. That preserves useful ideas while preventing an available answer from silently rewriting the question.