Backlog Grooming

What actually happens in refinement

A useful session does four things: clarifies items that are too vague to estimate, splits ones that are too large, reorders based on what has been learned since, and removes work that is no longer worth doing. The last of those is the one teams skip, and it is why backlogs grow into archives.

What a neglected backlog costs

The costs are indirect, which is why they are tolerated for so long.

Duplicates

The same work described three ways from three different meetings, each estimated separately, at least two of them eventually built.

Stale context

A ticket written eight months ago against a system that has since changed. Following it faithfully produces the wrong thing.

Planning friction

If the top of the backlog is not ready, sprint planning becomes a scoping workshop and takes three times as long.

Loss of trust

Once nobody believes the backlog reflects priorities, work starts arriving through side channels instead.

Keeping it from getting there

The habits that pay off are unglamorous: write acceptance criteria at creation rather than at planning, close anything you would not start this quarter instead of leaving it "for later", and check new items against existing ones before adding them. Duplicate detection at the point of creation is worth more than any amount of periodic cleanup, because a duplicate that never enters the backlog costs nothing to remove.