Why You Do Not Finish Projects
Many starts, few finishes? Why missing definitions of "done" block progress, and how clearly defined units of action change that.
"I start, but never follow through"
This pattern describes repeated starts without completion. Tasks are begun with energy but rarely fully finished, usually because no clear end state was ever defined. The solution is radically shrinking a task down to a clearly bounded unit with a predefined, verifiable completion criterion.
How the Pattern Shows Up
Many parallel projects and shifting priorities emerge. A task is started but replaced by a new one as soon as initial motivation fades, leaving the process open and incomplete.
What Is Really Happening
A precise definition of when a task is truly "done" is often missing. Without that definition, completion keeps getting pushed back because further improvements always seem possible, and the task fades out rather than being finished.
The Central Diagnostic Question
How exactly do you know that this task is done?
The Intervention
Shrink the task radically
Instead of a large project, a clearly bounded action that can be completed in a single pass is defined.
Define completion in advance
Before work begins, it is set concretely and verifiably what the finished state looks like.
Accept the endpoint as binding
Once the defined state is reached, the task counts as complete, regardless of the impulse to keep optimizing.
Example From Practice
A person regularly starts job applications, further training, and organizational tasks without ever clearly finishing them, a resume, for instance, keeps getting revised without ever counting as done. After committing to a single, shrunken task with a clear completion criterion, a full cycle of start, execution, and completion happens again for the first time.
Related Pattern
"I am waiting for the perfect opportunity"
Frequently Asked Questions
Why do I not finish projects?
Usually because a clear, verifiable definition of "done" is missing. Without a defined completion state, finishing keeps getting pushed back because further improvements always seem possible.
How do I define a good first step?
Small enough to be fully completed within a clearly bounded period, with a predefined, unambiguously verifiable end state.