The most common answer to "which lists do you keep?" is "one"

And that one list then holds “nail for the picture in the kitchen” right next to “plan family holiday” and “company antivirus rollout”. Plus an entry called “tax” with no further hint of what was meant.

Look at a list like that and you cannot start. Not because it is too much, but because every entry demands a different kind of thinking. The nail needs a hammer and four minutes. The holiday needs an hour of thinking with your partner. “Tax” first needs a decision about what it even refers to. Your head bounces between these demands and ends up doing nothing.

The answer is not sorting by priority. The answer is that each list answers exactly one question.

The question is the organising principle

Every list discussed here can be defined by the question you open it for. If you cannot say which question a list answers, you do not need it.

List	Question
Calendar	What has to happen today at a particular time?
Next actions, by context	What can I do now, here, with this energy?
Waiting for	What am I waiting on, from whom, since when?
Projects	What have I committed to?
Agenda, per person	What do I want to raise the next time we speak?
Someday/maybe	What do I not want to forget but not do now?
Reference	What do I want to find again when I need it?

Seven questions, seven places. The rest of this article covers what goes wrong with them in practice.

Calendar: the hard landscape

The calendar takes only what has to happen on a particular day or at a particular time. Appointments, flights, day-bound information.

Tasks do not belong in it, because the calendar knows nothing about backlog. Whatever did not get done in a past time slot disappears unnoticed instead of staying open. That is why people with full calendars swipe away their own reminders.

Next actions: context is more than a place

This is where the concrete physical actions live, sorted by what you need in order to carry them out. It is the list you keep open in working mode, and the only one holding something you can do right now.

Two things about it that most guides leave out.

The phrasing decides whether it is usable. “Quote for Miller” is not a next action, it is a topic. If a topic sits there, you redo the thinking every time you look at it — the thinking you should have done when you captured it. “Transfer Miller’s costing from the sheet into the proposal template” costs twenty seconds more when capturing and saves five minutes of run-up at every glance.

The classic contexts no longer all work. @Computer made sense when there was one computer and it stood on a desk. Today @Computer is a list holding ninety per cent of all tasks, and therefore worthless.

What holds up in practice: @Calls works unchanged, because it has a mode of its own. @Out and @Errands work, because they are tied to a place. @Home works. And the most useful cut today is not the place but the capacity available. One list for things that need full concentration, and one for everything that also works with half a brain on a train. Tired in the afternoon, you open the second one and still get work done.

Waiting for: the list that is almost always missing

Everything you delegated or requested and are still accountable for. With a date from which you have been waiting.

This list is missing in roughly every second system I see, and it is the most common cause of unpleasant surprises. Without a waiting-for list you remember an open commitment only when a client asks. And you then delegate less, because experience tells you delegated things vanish.

The date is not decoration. It is the only piece of information that turns an entry into an action. Something open for four days you leave running. Something open for three weeks is a phone call.

Projects: anything needing more than one action

A project is an outcome requiring more than one action. In GTD that is a low threshold, which is why thirty to a hundred open projects is normal with a full working and private life.

Nothing on this list is something you can do. It answers not “what do I do now” but “do I have everything in view”. You ask that question once a week, not at nine in the morning.

Phrase entries as an outcome, not an activity. “Antivirus rollout” is open-ended. “New antivirus solution active on all 240 clients” has a state you can recognise when it arrives.

Agenda lists: the underrated lever

One list per person you deal with regularly, and one per recurring meeting. Everything that is not worth a call of its own but should be raised the next time you speak lands there.

This is the list with the best effort-to-benefit ratio in the whole system, and the one kept least often. Two effects: you interrupt people considerably less, because small things can wait, and you never again leave a standing meeting realising afterwards what you had meant to ask.

With managers this is often the point at which the system changes something noticeable for the first time.

Someday/maybe: the valve

Everything you might want to do at some point but not in the coming weeks. The language course, the renovation, the book project.

This list has two functions. It keeps the active project list honest, because then only what is running is running. And it lets you write ideas down without committing to them. That permission is exactly why people start capturing everything in the first place.

The condition: skim it once a week. A list nobody looks at is a graveyard, and your head notices that soon enough and stops trusting it with anything.

Reference: everything with no action

Information where nothing needs doing but which you want to find again. Credentials, addresses, notes from meetings, contract documents.

This is also where my favourite among the minor lists belongs: the gift list. When somebody mentions a wish in passing, it goes straight in there, with the name. Before the next birthday that replaces half an hour of pondering with a glance. It is not a project and not a task, it is material that gets used exactly once a year and is then worth gold.

The boundary matters, because this is where everything tends to slide together. A list of films you want to watch is reference, as long as it is only a supply for the next film night. A particular book you want to have read by the summer is not reference but an intention, and belongs on someday/maybe or in the projects. The test is always the same: do I expect something to happen here.

The one test for everything

When you are unsure while processing where an entry belongs, three questions in this order will do:

Do I have to do something? If no, it is reference or someday/maybe or rubbish.

If yes: am I the one doing it? If no, it belongs on waiting for.

If yes: is one action enough? If no, it becomes a project, and the one action that gets it moving goes onto a context list.

That is the entire logic. Everything else is a tooling question.

Where it typically fails

Too many context lists. Create twelve contexts and you look at none. Four to six is almost always enough.

Lists in the tool, but the phrasings are topics rather than actions. The system then looks tidy and paralyses you anyway.

No weekly review. The lists drift so far from reality within two weeks that you no longer trust them. From that point you are carrying things in your head again.

Forcing everything into one tool. It is entirely fine for reference material to live in Obsidian or Drive and the next-actions lists in a task app. What has to hold together is the logic, not the software.

Common questions

Which lists do I need for GTD?

Calendar, next actions by context, waiting for, projects, agenda lists per person, someday/maybe and reference material. Each answers exactly one question, and that is precisely why they work better apart than together.

What is the difference between the project list and the next-actions list?

The project list holds desired outcomes and serves as a completeness check during the weekly review. The next-actions lists hold concrete actions and serve execution in daily work. Nothing on the project list is something you can do right away.

Do context lists based on places still make sense?

Partly. @Calls, @Out and @Home still work; @Computer is usually too big to help today. The more useful distinction is often the concentration available rather than the place.

How many context lists should I keep?

Four to six. Beyond that, the likelihood of looking at all of them regularly drops, and then you lose exactly the trust you were trying to build.

---

Count how many of the seven lists you actually keep. For most people waiting for and the agenda lists are missing, and those are precisely the two that change something fastest.