User stories

User stories must represent the business value. That's why we use the well known one-line description 'as an <actor> I want an <action>, so I can reach a <goal>'. It is both simple and powerful because it provides the team a concrete customer related context for identifying the relevant tasks to reach the required goal.

The stories pulled into the sprint by the team have a constraint on size. They should at least be small enough to fit into a sprint. This constraint of story size can in some cases require the story to be broken down into smaller stories. There are some useful patterns to do this like workflow steps, business rule or data variation etc.

Complex systems

When dealing with large and complex systems consisting of many interacting components the process of breaking down can impose problems even when following the standard guidelines. Especially when breaking down a story leads to stories which are related to components deep within the system without direct connection to the end user or the business goal. Those stories are usually inherently technical and far away from the business perspective.

Lets say the team encounters a story like ‘As a user I want something really complex that doesn’t fit in a story, so I can do A’. The story requires interaction of multiple components so the team breaks it down in to smaller stories like ‘As a user I want component X to expose operation Y, so I can do A’. There should be a user and business goal, but the action has no direct relation to either of them. It provides no meaningful context for this particular story and it just doesn't feel right.

Constraint by time and with no apparent solution provided by known patterns the team is likely to define a story like: ‘Implement operation Y in component X’, which is basically a compound task description and provides no context at all.

Components as actors

Breaking the rules a bit it is possible to use the principle of user story definition and provide meaningful context in these cases. The trick is to zoom into the system and define the sub stories on another level using a composite relation and making the components actors themselves with their own goals: ‘As component Z I want to call operation Y on component X, so I can do B’ and ‘As component X I want to implement operation Y, so I can do C’.

There is no direct customer or business value in this sub story, but because it is linked by composition it is quite easy to trace back to the business value. Each of the sub goals contributes to the goal stated in the composite story. Goal A will be reached by reaching both goal B and goal C (A = B + C).

Linking the stories

There are several ways to link the stories to their composite story. You can number stories like 1, 1a, 1b, ... or by indenting the sticky notes with sub user stories on the scrum board to visualize the relationship. To make the relation more explicit you can also extend the story description like: As an <Actor> I want <Action>, so I can reach <Goal> and contribute to <Composite Goal>.

Composite Stories


The emphasis of this approach is to try to maintain meaningful context while splitting (technical) user stories for complex systems with many interacting components. By viewing components as actors with their own goals you can create meaningful user stories with relevant contexts. The use of a composite structure creates logical relations between the stories in the composition and connects them to the business value. This way the team can maintain a consistent way of expressing functionality using user stories.


This method should only be applied when splitting user stories using the standard patterns is not possible. For instance it does not provide an answer to the rule that each story should deliver value to the end user. It is likely that more than one sprint is needed to deliver a composite story.  Also you should ask yourself the question why there is complexity in the system and could it be avoided. But for teams facing this complexity and the challenge to split the stories today this method can be the lesser of two evils.