The agile software development team holds a planning meeting at the beginning of each iteration to identify the stories that will be developed, and to break each of them down into specific technical tasks and acceptance criteria. Iteration planning meetings generally last from 2-8 hours, depending on the length of the iteration.  The longer the iteration, the more time needed to plan its details. The result is an iteration backlog of stories, defects, tasks, and acceptance tests.  Ideally, each iteration has a specified goal the team is attempting to achieve.

As an example, an iteration planning meeting for a two-week iteration is typically time-boxed to approximately four hours.  One approach many teams take is to separate the meeting into two sections, each with a different objectives.

During the first part of the meeting, the Product Owner (or product manager, or whomever on the team represents the domain expertise) presents the stories they are considering for the iteration to the team.  Each story is discussed to clarify its meaning and scope.  Larger stories are broken down and estimated as necessary.  Existing story estimates may be adjusted if during story clarification, new information comes to light that changes them significantly.  The team then selects the stories they feel they can complete within the iteration. Also during this first part of the meeting, the team will generally define the iteration goal or theme and assess their capacity in terms of available days or hours.  This provides another validation to ensure the team is not over-committed.

The second part of the iteration planning meeting takes the candidate iteration backlog and breaks it down into lower level execution details.  The design of the items in the iteration backlog is extended by decomposing the stories into tasks and clearly-defined acceptance criteria.  The team estimates each task, typically in hours or ideal days.  Once this decomposition analysis and design is complete, team members confirm that the estimates they have identified do not exceed the team’s anticipated availability/capacity for the iteration.  With that understanding, the team can commit to the iteration backlog and the iteration begins.