How to Prioritise Tasks: A GTD Decision Guide

The best way to prioritise tasks is the one you can run in thirty seconds, standing in a corridor, with your phone in one hand and a call starting in twelve minutes. The honest conclusion, up front: prioritising is not where the problem is. Almost every list that resists prioritising is a list that was never clarified, and no grid, matrix or ranking scheme can sort items that are not yet actions.

So this guide does two things. It shows you how to prioritise tasks using the four filters from David Allen’s Getting Things Done, in the order they actually work, and it shows you the setup in your task manager that makes those filters do the deciding instead of you.

Why Urgency and Importance Grids Stall on a Real List

Open your task manager now and read the first ten items. If three of them read like “sort out the website”, “Q3 reporting”or “Dave’s thing”, you do not have a prioritisation problem. You have a list of project titles, and a project title cannot be started, only worried about.

Try ranking “website redesign”against “email Sarah the three homepage headings”. One is an outcome that might take four months. The other takes ninety seconds. Urgency and importance are the wrong axes for that comparison, because the items are not the same kind of thing. The Eisenhower matrix assumes a clean, complete, single list of comparable actions. Most people have a half-list in a task manager, a second one buried in starred emails, a third in Slack messages they left unread on purpose, and a fourth in their head at 03:00.

Then there is the cost of deciding twice. An ambiguous item sits at the top of your list and you re-read it at 09:00 on Monday, again on Tuesday, again on Wednesday, each time making the same half-decision and moving on. That is not laziness. The item is genuinely undecidable in the four seconds you gave it, so your brain quite sensibly defers. Do that for a fortnight and you stop trusting the list, which is the real failure. Trust leaks out slowly, and the shadow list on paper appears.

Four things have to be true before any prioritising method works at all: everything is captured into one trusted inbox, each item has been clarified into a physical next action, every active project has exactly one available next action, and the stuff you are waiting on somebody else for lives somewhere other than your action list.

Clarify Before You Sort: Turning Open Loops Into Next Actions

The clarifying question is unglamorous and does nearly all the work. Is this actionable? If yes, what is the very next physical, visible action? Not “progress the website”, but “open the staging site and write the three homepage headings into notes2.txt”. You should be able to picture your hands doing it.

The two-minute rule is usually sold as a productivity trick. It is better understood as a prioritising tool: anything you can finish in under two minutes never enters the queue at all. Do it while clarifying and it is gone. That is why a clarified inbox is shorter than a captured one, and why the remaining items are all things that genuinely need a decision later.

When an item is really a project with no obvious first step, run Allen’s natural planning model over it rather than staring harder. “Sort out the website”becomes: purpose (stop losing enquiries from the contact page), outcome (new homepage and working form live by the end of next month), brainstorm (headings, photos, form provider, who signs off), organise (sequence those), next action (email Sarah the three homepage headings for approval). One stuck item, now one startable task and a project plan sitting behind it. The reference material for all that belongs in your notes app, not in the action list.

Waiting-for items are not priorities and should not compete with your own work. Delegated work goes on a separate Waiting For list with the date you handed it over and the person’s name, and you look at it once a week. That matters more than it sounds if you work across time zones: a question sent to Sydney on Thursday afternoon UK time is not blocked, it is simply asleep, and it should not sit in your Today view accusing you of something.

The Four Filters: How to Prioritise Tasks in the Right Order

Here is the part most guides skip. In GTD, priority is the fourth decision, not the first. Three cheaper filters run before it, and they eliminate most of the list without any judgement at all.

Context first. What can you physically do, here, with the tools and people available? At a client site with no VPN, half your list is impossible. On a train with patchy signal, the calls are out and the offline writing is in. Context is not a label you add for tidiness. It is the cheapest filter you own because it needs no thought.

Time available second. You have twelve minutes before a video call. Work out what fits. On a sixty-item list, context and time together will typically strike out the overwhelming majority in one pass, leaving perhaps three short calls, a two-line reply and one errand to check. That is a decision made in seconds instead of a full reshuffle. Notice that importance never came into it.

Energy third. Scheduling the hard, high-focus writing for 16:30 does not produce hard, high-focus writing. It produces a draft you rewrite on Thursday. Keep a Low Energy list, filed under the same contexts: expenses, filing, cancelling the subscription you forgot about, updating the project list. That list rescues the dead hour instead of letting you fake-work through email.

Priority last, applied to the handful of items that survived, judged against your current commitments rather than a static A/B/C label you assigned three weeks ago. Labels go stale within days. Your sense of what matters this week, refreshed in a weekly review, does not.

Run in this order, prioritising feels accurate at 15:40 on a Thursday. Run it the other way, starting with importance, and you end up choosing between four things you cannot do from where you are sitting.

Setting Up Views That Make the Decision for You

Availability settings do more prioritising work than any ranking scheme. If deferred, blocked and waiting-for items can appear in your main view, you are re-deciding on the same items every single morning. Fix the view and the decision shrinks.

In OmniFocus, four perspectives cover it: Today, Next Actions, Waiting For and Low Energy. Build Next Actions with availability set to first available action only, grouped by context, with completed and deferred items hidden. Custom perspectives need the Pro licence at the time of writing, so check the current pricing before you plan around it. I have written up the filter and grouping choices in more detail in this guide to building OmniFocus perspectives that cut your daily review, and there is an older, scruffier walkthrough in tricking out your OmniFocus perspectives.

Grouping matters more than people expect. Group by context for errands and calls, so you can clear five of them in one go. Group by project for deep work, so you see the sequence. Use a flat, ungrouped list for triage, because grouping slows down a pass where you are only making keep-or-kill decisions.

Add a Stalled Projects view and look at it weekly. Projects with no available next action, not overdue tasks, are the real source of that nagging sense that your priorities are wrong. Overdue items announce themselves. A stalled project just sits there radiating low-grade guilt, and a saved view surfaces the lot in seconds.

Things and Todoist do the same job more simply. In Things, tag contexts and filter the Today and Anytime lists by tag; the Someday list and start dates handle availability. In Todoist, build saved filters such as @calls &!subtask &!@waiting, though free plans have limited filters, so check the current allowance. None of this requires switching apps. If you are choosing a stack rather than fixing one, the rundown of Mac productivity apps that build a working GTD system covers what each tool is actually good at.

One more thing: triage email into your system rather than prioritising inside it. An inbox sorted by arrival time is a priority list written by other people, and it is louder than yours. Two-minute replies go now, actions get clarified into the task manager with a link back to the thread, reference gets filed, the rest is archived.

Horizons of Focus: Deciding What Matters, Not Just What’s Next

When two tasks both look urgent and the filters have not separated them, the tie-break is not a bigger grid. It is altitude. Allen’s horizons of focus run from current actions up through projects, areas of responsibility, goals and purpose. Ask which area of responsibility each task serves, and one of them usually turns out to belong to somebody else’s agenda.

Priorities should be reset in the weekly review and almost nowhere else. A realistic agenda, in a realistic timebox of around an hour: empty every inbox, check each active project for one available next action, scan Waiting For and chase what has gone quiet, then read the areas of responsibility list once. Do that and daily prioritising becomes a short choice rather than a full re-plan.

Renegotiate out loud. Agreements you quietly break keep resurfacing, because some part of you is still tracking them. Telling a colleague in Toronto that their report moves to next Tuesday takes one message and clears the loop properly. If you find yourself reshuffling the entire list every morning, that is not diligence. It is the symptom of a skipped review, and it is also where most people’s productivity system quietly stops working.

Where AI Helps With Prioritising, and Where It Makes It Worse

AI is genuinely useful at three jobs here. Breaking a vague open loop into candidate next actions, so “sort out the website”comes back as eight plausible first steps you can accept or reject. Drafting a project outline as support material. Summarising a forty-message thread before you triage it, which saves real minutes on a Monday.

It is bad at the one thing people keep asking it to do. A model cannot decide which commitment matters to you, because it does not know what you promised, to whom, or what you will be embarrassed about in March. Ask it to assign priorities and it will. Ask it for deadlines and it will invent them. Fake urgency enters the list, you notice within a week that the top item is nonsense, and trust leaks out again.

Prompt for actions, not advice. “Rewrite each of these as a physical next action starting with a verb, under eight words, no dates”produces something usable. “How should I prioritise my week?”produces an essay. Then review every suggestion before it enters the list, with the same clarifying question you would apply to your own capture. Keep the rest as project support material. More detail on which jobs hold up and which do not is in this breakdown of how to make ChatGPT work with GTD, and on catching the thought in the first place, ubiquitous capture with an AI assistant.

Common Prioritising Mistakes and How to Fix Them This Week

Flagging everything. A flag is a signal, and a signal on forty items is decoration. Cap flags at what you intend to finish today and clear them.

Treating due dates as priorities. Reserve due dates for the hard landscape: things that genuinely break on that date, like a filing deadline or a flight. Everything else gets a defer date or no date at all. Use due dates as a nudge and you train yourself to ignore the overdue badge, which destroys the one alert that should mean something.

Prioritising inside a tool you do not trust. If a paper list lives next to your keyboard, the paper list is the real system. Find out what the app failed at (usually capture speed or a view that shows unavailable items) and fix that before ranking anything.

If it has all slipped, the recovery sequence is short. Capture everything into one inbox, including the notebook and the starred emails. Clarify each item to a next action, doing the two-minute ones as you go. Build one context view with availability filters on. Review weekly. That is the whole of how to prioritise tasks reliably: clarify first, then let context, time and energy do the elimination, and save your judgement for the three or four items still standing.

Frequently Asked Questions

What is the fastest way to prioritise tasks when everything feels urgent?

Filter before you judge. Ask what you can physically do where you are, how long you have, and what your energy is like, then look only at what survives. Everything feeling urgent is usually a sign that unclarified project titles are sitting in the list pretending to be actions.

Should I prioritise by due date or by importance?

Neither on its own. Due dates belong to the hard landscape, meaning things that genuinely break on that date, and importance is the last filter rather than the first. If most of your items carry a due date, the dates have become wishes and you will start ignoring them.

How do I prioritise tasks when my list is spread across email, chat and notebooks?

You cannot, and no method will fix it while the list is fragmented. Collect everything into one inbox, clarify it into next actions, and set up email triage so the inbox feeds the system rather than competing with it. One trusted list is the precondition, not a nice-to-have.

Does the Eisenhower matrix work with GTD?

It works at the end, on a small set of already-clarified actions, which is roughly where GTD’s fourth filter sits anyway. It does not work as a first move on a raw list of outcomes and half-decisions. Clarify, filter by context, time and energy, then use whatever framework you like on the remainder.

How often should I review and reset my priorities?

Once a week, properly, in a review of about an hour: inboxes to zero, every project checked for one available next action, Waiting For scanned, areas of responsibility read once. Daily, you are choosing from an already-trusted list, not re-planning. A daily full reshuffle means the weekly review was skipped.

Can ChatGPT help me prioritise my task list?

It can help you get the list into a prioritisable state by turning vague loops into candidate next actions and summarising long threads before triage. It cannot decide what matters to you, and letting it set priorities or deadlines imports urgency nobody agreed to. Treat its output as raw capture and clarify it yourself before anything enters the list.

If you want the next piece of this, read the perspectives guide and build the Next Actions view with availability filters on. That one change removes more daily deciding than any ranking method will.