A task is a bounded unit of work that can be assigned, attempted and judged complete.
Writing an article, checking a calculation, repairing a pump, calling a parent, approving a budget, backing up a database and preparing for an examination are all tasks. They differ in purpose, effort, urgency, dependency, recurrence, ownership, uncertainty and completion criteria.
Quick answer: how should tasks be categorised?
- Purpose: what outcome does the task contribute to?
- Effort: how much work or capacity is required?
- Priority: how important is it relative to alternatives?
- Deadline: when must it be completed?
- Dependency: what must happen before or after it?
- Recurrence: one-off, periodic, event-triggered?
- Ownership: who is responsible for completion?
- Risk: what happens if it fails or is delayed?
- Automation: human, machine or hybrid?
- Completion: what evidence proves it is done?
This article applies the framework from How to Categorise Anything to tasks as operational units of work.
1. Define the task boundary
“Improve the website” is usually too broad. “Update the article title and verify the link” is closer to a bounded task.
2. Separate task from project
A project contains many tasks coordinated toward a larger temporary goal.
3. Separate task from process
A recurring process may generate many task instances over time.
4. Purpose connects task to outcome
Every meaningful task should support a result, decision, process or capability.
5. Output should be explicit
Tasks are easier to classify when the expected deliverable or state change is known.
6. Completion criteria matter
“Done” should mean something observable: file published, test passed, item delivered, approval recorded.
7. Effort is not duration
A task can require two hours of work spread across three days of waiting.
8. Duration is not deadline
A ten-minute task due next month differs from a ten-minute task due in five minutes.
9. Urgency is not importance
An urgent task may have low strategic value, while an important task may have a distant deadline.
10. Priority combines context
Priority may depend on value, urgency, dependency, risk, effort and commitments.
11. Priority labels need rules
High, medium and low are useful only when people understand how they are assigned.
12. Dependencies create ordering
A task may be blocked until evidence, material, approval or another task arrives.
13. Predecessor and successor differ
Record what must happen before the task and what depends on it afterward.
14. Critical-path tasks deserve distinction
Delay to some tasks delays the whole project; delay to others may be absorbed by slack.
15. Recurring tasks need templates
Daily, weekly, monthly and annual tasks are best represented as one recurring definition plus individual occurrences.
16. Event-triggered tasks differ from schedules
Some work begins only when a threshold, message, incident or approval occurs.
17. Ownership should be singular enough for accountability
Many people can contribute, but one role should usually be accountable for completion.
18. Assignee and approver are different roles
The person doing the work may not have authority to accept the result.
19. Human and automated tasks differ
Automation can execute repeatable work, while judgement-heavy tasks may require human review.
20. Hybrid tasks contain handoffs
A model may draft, a human may review, and software may publish. Preserve each responsibility.
21. Risk belongs with the task
Failure impact, safety, compliance and reversibility can determine review level.
22. High-risk tasks need stronger controls
Deletion, financial release and safety-critical work require more evidence and approval than routine formatting.
23. Uncertainty affects planning
Unknown effort, unclear requirements and external dependencies should be recorded explicitly.
24. Research tasks differ from execution tasks
Research reduces uncertainty; execution applies an already chosen method.
25. Decision tasks create commitments
Some tasks culminate in approval, selection or judgement rather than a physical deliverable.
26. Communication tasks create shared state
Notify, brief, document and hand over are tasks whose outcome is changed information across people.
27. Maintenance tasks preserve capability
They may not create new output but prevent degradation.
28. Recovery tasks restore capability
Repair, rollback and replacement respond after failure.
29. Creative tasks have less deterministic outputs
Completion criteria should emphasise fit and quality rather than one predetermined exact result.
30. Administrative tasks coordinate systems
Scheduling, filing, approving and updating records may be low novelty but high operational importance.
31. Task state is separate from task type
Backlog, ready, blocked, in progress, review, done and cancelled describe workflow state.
32. Blocked needs a reason
Waiting for approval, data, material, person or system should be distinguished for resolution.
33. AI can classify task intent
Models can route work toward research, drafting, calculation, scheduling or review, but consequential execution still needs governed permissions.
34. AI can estimate effort imperfectly
Historical patterns may help, but novel tasks need uncertainty ranges rather than false precision.
35. Task taxonomies drift
New tools automate old work and create new review tasks. Categories should evolve with actual workflows.
36. Completion evidence should be stored
Links, receipts, test results, approvals and version IDs help prove work was completed correctly.
37. A practical task record
- task ID;
- purpose;
- output;
- completion criteria;
- effort;
- priority;
- deadline;
- dependencies;
- recurrence;
- owner;
- assignee;
- approver;
- risk;
- automation level;
- status;
- completion evidence;
- version.
38. Task categories should improve routing
Useful categories determine who should do the work, how much control is needed and what completion looks like.
39. Workload is a system property
Individual tasks become queues, bottlenecks and capacity problems when viewed together.
40. The deeper idea
A task is where intention becomes scheduled work with an accountable finish line.
To categorise a task well is to know why it exists, who owns it, what blocks it, when it matters and what evidence proves it is finished.
Final answer
Categorise tasks by purpose, output, effort, priority, deadline, dependencies, recurrence, ownership, risk, automation and completion criteria. Keep urgency separate from importance, state separate from type, and assignee separate from approver. Store completion evidence for consequential work.