What Goal Decomposition Really Is
Goal decomposition is the process of taking a high-level outcome and breaking it into progressively smaller executable units until each unit takes less than 2 hours and has a clear completion criterion.
The decomposition hierarchy for a well-structured goal: North Star → Milestones (monthly/quarterly checkpoints) → Weekly Targets → Daily Actions. Each layer must be specific enough to generate the next layer without ambiguity.
- ▸North star: the overarching outcome (6–12 month horizon)
- ▸Milestones: 3–6 checkpoints with measurable completion criteria
- ▸Weekly targets: the output needed each week to hit the next milestone
- ▸Daily actions: tasks taking under 2 hours, with clear "done" criteria
The AI-Assisted Decomposition Flow
LLMs are unusually good at goal decomposition because they can apply domain knowledge (what typically has to happen to achieve Y in domain Z) combined with constraint awareness (timeline, available hours, dependency ordering).
An effective AI decomposition prompt includes: the goal, the timeline, your current status, your available weekly hours, and any known constraints. The output should be a milestone-and-task structure, not a flat list.
- ▸Include timeline and weekly capacity in your AI prompt
- ▸Ask for dependencies explicitly: "What must be completed before X?"
- ▸Request a "first 7 days" action set to give you immediate traction
- ▸Refine iteratively: "Reduce week 1 scope by 30% for a sustainable start"
The Dependency Tree Problem
Most goal decomposition fails by producing a flat list of tasks without dependency relationships. In reality, many tasks can't start until others finish. Missing this produces parallel actions that aren't executable concurrently.
Explicitly mapping dependencies prevents the experience of "I don't know what to do next" — you always have a clear critical path and can see which tasks are currently unblocked.
- ▸For each task, write: "This requires: [dependency] to be done first"
- ▸Map the critical path: the sequence whose delay delays the entire milestone
- ▸Identify which tasks can run in parallel vs. must be sequential
Maintaining the Decomposition Over Time
Goal decomposition isn't a one-time planning exercise. It's a living structure that requires weekly recalibration as reality diverges from the initial plan.
Treat your decomposed plan as a hypothesis. Weekly: what completed, what slipped, what changed in priority. Monthly: does the milestone architecture still reflect the current best path to the north star?
- ▸Review decomposed tasks weekly, not just at milestone boundaries
- ▸Expect and plan for 20–30% re-planning overhead in complex goals
- ▸When scope changes, update the dependency tree, not just the task list