A next actions list stalls for one reason far more often than any other: the lines on it name a topic instead of a movement. "Taxes." "Boiler." "Mum." Each was clear when you wrote it and each costs you thirty seconds of re-deciding every time you read it, which is why the list gets skimmed instead of worked.
The fix is mechanical and takes about twenty minutes for a list of forty. Rewrite every line so it describes something your body does, somewhere specific, and the list stops being a set of reminders to think.
What a next actions list is for
It holds the single physical step you would take next on each live project, and nothing else. Not the project. Not the outcome. Not a reminder that the project exists.
The rule is strict because the list gets read at the worst moments — twelve minutes between calls, tired, on a phone. You have enough capacity there to do a thing and none at all to work out what the thing is. A line that needs a decision before it needs an action loses to your inbox every time.
Where the five GTD lists sit relative to each other, and which ones an app has to carry, is laid out in how to pick an app for GTD. This post is only about the wording.
Why physical wording beats sensible wording
There is real evidence behind this, and it is not about willpower. Peter Gollwitzer's 1999 paper in American Psychologist, Implementation intentions: strong effects of simple plans, starts from an uncomfortable finding: intentions predict behaviour, but weakly, accounting for only 20% to 30% of the variance in what people actually do. Meaning well is a poor predictor of acting.
What closes the gap is specifying the situation. Gollwitzer's format is "whenever situation x arises, I will initiate the goal-directed response y", and his argument is that such plans hand control of the response over to an anticipated cue, so that meeting the cue triggers the action without a fresh act of deliberation.
A next actions list is a rough, written version of the same idea. "Reply to Priya's email with the two dates" carries its own cue — it fires the moment you are in the mail client. "Sort out scheduling" carries nothing, so it needs you to be in a deciding mood, and deciding moods are rare on a Tuesday afternoon. Habit people have been borrowing the same mechanism for years by attaching a new behaviour to an existing cue, which is what habit stacking is.
Ten lines and their rewrites
This is a real pass over a real list. The middle column is what actually went wrong, because the fix is different in each case.
| what was written | why it stalls | the rewrite |
|---|---|---|
| taxes | a project wearing an action's clothes | download the annual statement from the bank portal |
| mum | no verb and no ending | call mum, Sunday after lunch |
| boiler | a noun standing in for a decision | call Greenway and ask for a service quote |
| sort out insurance | the verb is too abstract to picture | find last year's policy PDF in the mail archive |
| follow up with Priya | channel unspecified, so it needs a plan first | reply to Priya's 4 Sept email with the two dates |
| gym | an identity, not a movement | pack the kit bag and leave it by the door |
| plan the trip | four actions in a trench coat | pick the week in October and block the calendar |
| fix the website | unbounded, so it never looks finishable | replace the broken contact link on the about page |
| think about the new role | thinking has no physical form | write three lines on what would make it a yes |
| dentist | the outcome is missing | book a check-up through the practice's online form |
Read the middle column top to bottom and the same four faults keep appearing: no verb, a verb you cannot picture, a hidden project, a missing location. None of them needs judgment to spot, so the pass below can be done quickly and badly and still work.
The dentist row is worth sitting with. "Dentist" survives on lists for years because the actual work is finding the practice's booking page, and that friction is invisible in the word.
Rewrite your list in six passes
Do these as six separate sweeps rather than perfecting each line as you meet it. Six mechanical passes are faster and more consistent than forty judgment calls.
- Add a verb to every line that has none. Do not improve anything else yet. This pass alone usually catches a fifth of the list.
- Replace every abstract verb — sort, handle, deal with, look into, organise — with something physical: call, email, find, print, book, pack, write, open.
- Add the location or channel where the line is missing one. "Book the flights" becomes "book the flights on the airline site", which sounds redundant until you are looking at it on a phone in a queue.
- Split anything that hides more than one step. If the rewrite contains "and" between two different verbs, it is a project. Give it a project and keep only the first step on the list.
- Add dates to the ones that have real ones, and only those. An invented deadline teaches you to ignore deadlines, which is expensive because the real ones are in the same column.
- Read the whole list out loud, fast. Anything that makes you pause is still wrong. Pauses are reliable in a way that re-reading is not.
Twenty minutes for forty lines is normal on the first run, and about five minutes a week afterwards during the review. If you keep needing a full rewrite, the problem is upstream, at capture — see processing the GTD inbox for the pass that stops raw thoughts from landing on the actions list in the first place.
The two fields that decide whether the list is honest
A next actions list dies of false urgency more often than it dies of bad wording. Everything gets a due date, most of those dates were invented to make the item surface, and within a month the overdue count is thirty and means nothing.
The correction is to separate when you could start something from when it is actually due, and to leave the second field empty most of the time. Most tasks have a start date and no deadline. Treating those two as one field is the single most common reason a list becomes noise, and it gets a full post in start date vs deadline.
Lines that resist every rewrite
Some lines will not turn into an action no matter how many verbs you throw at them. "Learn German." "Move somewhere warmer." "Do something about the garden."
That resistance is information. It usually means the item is a decision you have not made, or something you want in general and have not committed to, and both of those belong on a someday maybe list rather than on the actions list where they get skipped weekly. The weekly pass that catches them is in the GTD weekly review checklist.
Where init.Tasks fits
init.Tasks runs keyboard-first on web and Mac: a terminal-looking home for to-dos, personal projects and the material behind them, on iOS, macOS and web. Quick add takes plain language, #project files an item and [tag] labels it, so the rewrite pass above is typing rather than form-filling.
Start date and deadline are genuinely separate fields, which is what lets a line stay honestly worded: a step can be startable from Monday without claiming to be due Monday, and the deadline column keeps meaning something. Notes and links attach to any to-do, so the booking URL sits on the line that needs it instead of in a browser tab you closed three weeks ago.
Saved filters cover contexts if you want them: one for calls, one for errands, one for anything tagged [waiting]. The AI connection is coming in september, and the use here is the boring one — asking a connected assistant to find every line with no verb in it. The beta waitlist is open and it is free while in beta. The feature list says what is built and what is not.
FAQ
What is a next action in GTD?
A next action is the single physical, visible step you would take next to move a project forward — something you could hand to someone else and have them do without asking a question. "Call the garage" is a next action. "Car" is a project, and "sort out the car" is a project with a verb glued on.
How many next actions should a list hold?
One per live project is the target, which for most people means forty to sixty lines. More than that usually means outcomes are being listed alongside steps. If the list feels unreadable at eighty items, the fix is not filtering — it is finding the twenty items that are actually projects and moving them.
What is the difference between a next actions list and an ordinary to-do list?
An ordinary list mixes topics, outcomes, reminders and steps in the same column, so reading it requires interpretation. A next actions list holds only steps, each tied to a project that lives elsewhere. The practical difference shows up when you are tired: one list can be worked in a spare ten minutes, the other has to be decoded first.
Should a next actions list use contexts?
Contexts — at the computer, on the phone, errands — mattered more when tools were physically separate. Most people now carry every context in one device, so a light version works better: two or three tags for genuinely distinct situations, usually errands, calls and anything blocked on another person. Twelve contexts is a filing system, and filing systems get abandoned.