Product backlog

Go back

The Product backlog, is a sort of project specification (cahier des charges). This is a list of tasks. You must add the features (~=user stories) that the client requested.

  • Usually, they are prioritized (MoSCoW)
  • Usually, they are sorted by importance (most important to less important)
  • Usually, you must add the dependencies between tasks

Note: The entries of the product backlog are also called items.
Note: Sometimes, the backlog may become too big, so you may remove the less important tasks, and label them as "out of scope" (such as the Could/Would, or tasks that do not impact the sprint goal if they are removed).
Note: reviewing the backlog is sometimes called backlog grooming (or refinement)
Note: Usually, the Product Owner and the Scrum Master are preparing the Product backlog before the meeting.

User Stories and Epics

An Epic is a group of (user) stories that are "similar" (ex: all stories related to the user management). A User Story is a text describing a task from the point of view of a user. You got stories about

  • features (functional)
  • bugs (technical)
  • tests (technical)
  • technical debt (technical): this is what you have to do before starting to improve the code (refactor/create constants/normalize things/...)
  • actions (technical): changing your way of doing things (=from a retrospective)

A user story is supposed to provide content about the task, so that you are "working with" the one that actually needs it. We must know the reason, the goal, the value, and an estimation of the cost/value for the person needing this task to be done.

The usual template is "As a <type of user>, I <want to do something/goal>, so that <a reason/benefit>.".

Note: to add a technical story, you need to explain and persuade the product owner that this will help
Note: you will usually deliver the stories when the epic is done.
Tip: give colors to your stories according to their kind (ie: feature, bugs, etc.).
Note: You can see an epic as the result of a succession of stories, and a group of epics is making an initiative (omitted).


You must filter the tasks by priorities, most likely using the MoSCoW notation.

  • M: Must have this
  • S: Should have this
  • C: Could have this
  • W: Would have this

You may ask the client about the priority and/or deduct it by yourself, since some tasks that the client wants may be too hard/useless if done in the first increment. You must take into account the value for the client.


You need to evaluate how much complex a task is (or how much effort it will require). We are not evaluating the duration like in non-agile projects.

  • You may compare tasks and sort them by difficulty.
  • You can give a value among a fixed list (0, 1/2, 1, 2, 3, 5, 8, 13, 20, 40, 100, ?, inf.) to each task. The average/median value may be used as an answer.

If you got outliers (valeurs abérantes) such as (almost) everyone gave "10", one gave "5", and one gave "20", then it would be a good idea to ask them why they gave this value (maybe they thought of something others didn't).

You should learn from your previous estimates. You may pick a previous task and use it as a reference.

The backlog itself

You may create a table, in which each row is a story. You may group them by epics if you want. As for the columns, here are some ideas

  • # or ID: ID of the story
  • Title or Title+Description: the story
  • The priority
  • The importance, estimated with
    • An estimation of the complexity
    • An estimation of the value for the client
  • Some notes
  • The Sprint: the one in which you completed this story (empty if you didn't)
  • The Epic: if this story belongs to an epic
  • The dependencies
  • How will the Product Owner (or ...) test that the feature is working (acceptance tests)