Getting Things Done: a complete guide to GTD

Allen published Getting Things Done: The Art of Stress-Free Productivity in 2001 and revised it in 2015. It is shelved as a productivity system, but its argument is cognitive: the mind is good at having ideas and bad at holding them, so anything agreed to but not recorded reliably becomes an open loop, a background process that demands attention without making progress.

The state Allen calls mind like water is what remains when there are no open loops. You are not calmer because you have less to do, but because nothing you have to do depends on remembering it.

This is why GTD works bottom-up. OKR and SMART goals start from what you want to achieve and work downward; Allen argues that fails while attention is consumed by the unprocessed layer beneath. GTD therefore does not tell you what matters: it handles volume and keeps agreements. If your problem is too many equally reasonable things to do, the Eisenhower Matrix is the tool you want, and GTD keeps the list it operates on honest.

The 2015 edition names the steps Capture, Clarify, Organize, Reflect and Engage; the first edition used collect, process, organize, review and do, but the workflow is identical. Most failed implementations fail at Clarify or Reflect, because those are the two steps that require sitting down and making decisions.

Capture means putting anything with your attention into a trusted collection point, immediately, without judging it: a task, a half-formed idea, a nagging worry, the thing you noticed about the invoice. Three properties make a capture tool work.

• It is always to hand. A tool you have to walk to loses items.
• It costs almost nothing to use. The moment capturing requires a project, a due date and a tag, you weigh whether a thought is worth the effort, and the small ones go unwritten. Capture is a dump, not a filing operation.
• It gets emptied. An inbox never processed teaches you the system is unreliable, and once you stop believing it you go back to holding things in your head.

Start with a full mind sweep: work through trigger categories (projects, people, commitments, home, finances, things that are bugging you) and write down every unresolved item until nothing is left. The first sweep usually produces well over a hundred items and takes a couple of hours. It is unpleasant, and it is the most useful thing in the method. Keep inboxes few afterwards but do not insist on one: email, a notes app and a paper tray are fine if all three empty on the same rhythm.

Clarify is where an inbox item stops being a vague thing and becomes something you can act on or safely ignore. Take one item at a time, off the top, and never put it back: Allen's rule is that nothing returns to the inbox once picked up. Non-actionable items go to trash, reference or someday/maybe, and most of an inbox turns out to be one of those three. Where an item is actionable, the output is the very next physical action: not "plan the offsite" but "email Priya for the venue shortlist."

Organize is putting each clarified result where it belongs. Allen uses a small fixed set of containers, and the discipline is in refusing to add more:

• Next actions. Single physical actions you can start now, usually split by context. The list you work from.
• Projects. A flat inventory of every outcome needing more than one step. Titles, not plans. Its job is to guarantee each has a next action somewhere.
• Waiting for. Things other people owe you, with the date you asked. This turns delegation from hope into tracking.
• Calendar. The hard landscape only: appointments and things that must happen on a given day. Nothing aspirational.
• Someday/maybe. Things you have decided not to decide yet.
• Reference. Everything with no action but future value. Here GTD hands off to a note system such as PARA or a Zettelkasten, which do this layer better than a task manager can.

The calendar rule deserves emphasis, because breaking it destroys trust fastest. Put a soft intention there ("Thursday: write the report"), let Thursday pass, and you have taught yourself that calendar entries are suggestions. Then the real appointments start looking like suggestions too. Keep intentions on the next-actions list, which is where time blocking complements GTD rather than replacing it.

Above these lists sit the horizons of focus, described as altitudes: current actions on the runway, then projects, areas of responsibility, one-to-two-year goals, three-to-five-year vision, and purpose. Almost everyone lives at the bottom two and visits the rest occasionally.

A context filters the next-actions list by asking: given where I am and what I have with me, what can I actually do? Allen's examples are @calls, @computer, @errands, @office, @home, @agendas and @read/review. In 2001 these were sharp: a phone was a device in a room, and leaving the office put most of your work out of reach.

The smartphone collapsed that. @calls, @computer and @online now describe the same place, which is everywhere, and a context that never excludes anything is not filtering. This is the commonest objection from people adopting GTD today, and it is fair. What survives is the underlying idea, which was never really about location: a context is any constraint that makes part of your list unavailable. The ones that still bind are different.

• Energy. A @braindead list (expenses, filing, triage) and a @deep list beat any location split, because what you cannot do at 4pm is hard thinking, not phone calls. This is the overlap with energy management and deep work.
• Time available. A @15min list gets the gaps between meetings used instead of skipped. An action you cannot start in the time you have is deferred until it is urgent.
• People. @agendas, per person or per recurring meeting, has aged best. Every "I should mention this to Sam" lands on Sam's list instead of interrupting you, and the one-to-one agenda assembles itself.
• Tools and places that still constrain. @errands is intact, as is anything needing a specific machine, a physical document, or the work laptop's VPN.

Keep the number small: a context you never filter by is a tag maintained for nothing, and a list split eleven ways is one you stop opening.

Allen calls the weekly review the critical success factor, and it is not one habit among five: it is the habit the other four depend on. Capture, clarify and organize produce a system. Only the review keeps it true, and a system you suspect is stale is one you work around within a fortnight. The book structures it in three passes:

• Get clear. Empty every inbox to zero, then sweep for anything held only in your head.
• Get current. Review the calendar backwards over the past week or two, then forwards over the next few. Walk the next-actions list and cross off what is done. Walk the projects list and confirm each has a next action. Chase the waiting-for list.
• Get creative. Read someday/maybe. Promote what has become real, delete what you will never do.

Budget an hour or two at first; it gets faster as the backlog shrinks. Fix it to a slot, because the same hour every week is far more durable than doing it when you feel like it.

The backward calendar pass is the highest-yield part for anyone whose commitments arrive in conversation, since agreements accumulate faster than note-taking. If that is your week, a decision log and a post-meeting workflow do much of this before the review reaches it.

Allen spends the least time on the final step, deliberately: if the first four are done, the fifth is mostly obvious. He offers three models. The first is a four-criteria filter applied in order: context, time available, energy available, then priority. Priority comes last because it is the only judgement call, and judgement is far easier across five candidates than across two hundred.

The second is the threefold nature of daily work: at any moment you are doing predefined work from your lists, doing work as it shows up, or defining work by processing inputs. All three are legitimate. The failure is doing only the second, which feels productive and leaves the system to rot. The third is the horizons as a review ladder: when the lists feel wrong, the answer is usually a level up.

GTD is otherwise agnostic about execution. It pairs naturally with Pomodoro, which supplies the how once GTD has answered the what, and with the Ivy Lee method for cutting a long list to a daily shortlist.

• An inbox that is never emptied. The commonest failure by a wide margin. Capture is enjoyable and clarifying is work, so the inbox becomes a second, worse to-do list, and because unprocessed things hide in it you stop trusting the processed lists too. Emptying does not mean doing; it means deciding.
• Next actions that are really projects. "Website redesign" is not a next action, and every time you read it you re-experience the planning you have not done. That friction is why items sit untouched for months. The test is physical: could you start it in five minutes without thinking?
• Skipping the weekly review. Miss one and the system is slightly stale. Miss three and you no longer know whether the lists are true, so you keep the real priorities in your head as a hedge, the exact state GTD exists to prevent. A fifteen-minute version beats none.
• Building the system instead of running it. Migrating between apps, redesigning the tag taxonomy and reading more about the method all feel like progress and produce none. Allen's system is content-agnostic on purpose.
• Treating someday/maybe as a graveyard. A list nobody reads is a deleted list that also makes you feel worse. Read it in the review and delete freely.

One clarification that trips people up: Allen's two-minute rule and James Clear's are unrelated. Allen's is a processing shortcut, finish it rather than track it. Clear's, in Atomic Habits, scales a habit down to a two-minute version so that it starts at all.