What a good task looks like
The short answer
A task is well written when you can start it without deciding anything first.
Four properties, and most badly-behaved items fail the first:
- It names a first physical action. Draft the intro paragraph, not sort out the report.
- It is estimable. If you cannot put a rough time on it, it is not understood yet.
- It has one finish. You can say whether it is done without interpreting.
- It is one task. If it contains an "and", it is two.
The five-second test: say the first thirty seconds out loud. If you cannot, the wording is the problem — not your discipline, and not the priority.
Why wording matters more than it sounds
A vague task makes you do two jobs at the moment you look at it: work out what it means, then do it. The first job has no obvious start and no visible progress, so it is the one that gets deferred — and the whole item stalls behind it.
This is why a task can survive weeks of being scheduled later while everything around it moves. Nothing was wrong with its priority. There was simply nothing to begin.
The four properties, concretely
A first physical action. The test is whether it names something your hands do. Email Priya asking for Q3 figures passes. Follow up with Priya does not — it still requires deciding what "follow up" means, and that decision is the obstacle.
Estimable. Not accurate: estimable. If no rough figure comes to mind, you do not yet know what the task involves, and find out what this involves is the real first item. This is also what makes a day's arithmetic possible at all — what an estimate is actually for.
One finish. Improve onboarding has no end state, so it can absorb unlimited time and never be complete. Cut the signup form to four fields ends.
One task. "Write the deck and send it to legal" is two items with a dependency, and writing them as one hides the dependency — which is where blocked work goes to become invisible.
What to do with the bad ones
Do not delete them. They are usually pointing at something real.
Rewrite as the first action. Sort out the website → list the three pages that need changing. That is startable, estimable and finishable, and it will produce the next real task.
Split on the "and". Two items, and note which blocks which.
If you cannot name a first action, make that the task. Spend twenty minutes working out what the migration actually involves is a perfectly good item. It is honest about being investigation rather than execution, and it ends.
Delete the ones that are wishes. Be better at email is not a task. Either it becomes a specific action or it is an intention, and an intention on a task list costs attention every read and returns nothing.
The one exception
Not everything needs this much precision. A list of small, obvious, same-kind items — a shopping list, a batch of expenses — works fine as fragments, because the first action is self-evident.
Precision is worth the effort exactly where things stall: the item you have looked at four times and not started. That is where the wording is almost always the problem, and where a ninety-second rewrite outperforms another day of resolve.
Where to go next
- What to do when you can't start — when the first action is the whole obstacle
- How long should a task take? — estimability, and what an estimate is for
- What to do with a task you keep postponing — the four things it might be
Using this in Decide One
Decide One asks for a duration on every line, which quietly enforces most of this: a task you cannot estimate is a task you cannot enter properly, and that is usually the first sign it needs rewriting rather than rescheduling.
A line you did not finish is unfinished. It is not marked against you, and a day you did not open is not counted.
Open Decide One — free to use, with no account. Nothing you write leaves your device.
Decide One is a priority instrument. Bring the day into view, choose a method, give the first thing real time.