Some commit to their work and complete requirements throughout -- not just at the end.
Other teams struggle. Their sub-tasks may make progress, but their overall requirements or "stories," which express requirements in ways that customers can relate to, seem to get stuck. They finish on the last day of the iteration, if at all.
What makes these teams different?
Often requirements haven't been sub-divided. Queuing theory teaches that the same amount of work divided into smaller pieces flows faster. Teams with stories divided into work durations of one to three days see their work fly through the system. They can finish some requirements and then pick more.
Teams with stories that take a week or more are at risk of a traffic jam. Moreover, we're less aware of the delay until later -- when it's harder to take corrective action.
One correction is to refocus on a smaller number of requirements, but dedicate to finishing those. Another method is to split a story, even though the iteration is underway. Or, remove a story from the current iteration so it can be fully completed in another.
If none of these ideas seem enough, make sure the team is committed. Per the principles in the Agile Manifesto, team members need to self-organize to dedicate themselves to finishing whatever work is planned.
How have you avoided Agile traffic jams in your projects? Has splitting stories to a manageable size helped avoid bottlenecks?