How to plan your week when half of it is already booked

Most advice on planning a week assumes an empty week. Sit down on Sunday, choose three priorities, allocate the days. That works for someone whose calendar is a blank grid. For anyone whose Monday already contains four meetings booked by other people, the exercise is not allocation. It is negotiation with a week that has already been partly spent, and the planning session that ignores this produces a list that Tuesday afternoon destroys.

The useful version starts from what is already committed, not from what would be ideal.

Start by counting what is left

Before deciding what the week is for, the honest first question is how many hours are actually available. That number is almost always smaller than it feels, and the gap between the felt number and the real one is where plans die.

Three subtractions do most of the work. Meetings already accepted. Movement between anything that happens in person. And the fragments, meaning gaps of twenty or thirty minutes between commitments, which are real time and are not usable for anything that requires picking up a thread.

The fragments are the subtraction people refuse to make. A Tuesday with four meetings scattered across it can show three hours of white space and contain no single usable block. A Tuesday with four consecutive meetings and the same total shows the same three hours and contains one. Those two days are not equivalent, and no total of free hours can tell them apart.

So the first output of a planning session is not a list of intentions. It is a count of usable blocks. Two blocks of ninety minutes is a week that can carry one piece of substantial work. Six fragments is a week that can carry correspondence, review and decisions, and nothing else. Both are legitimate weeks. Confusing one for the other is what makes planning feel like a lie by Wednesday.

Put the commitments on the grid before choosing the work

A plan can only be as accurate as the calendar it is drawn from, and most calendars are missing two kinds of entry that have every right to be there.

The first is travel. Time between an in-person appointment and whatever follows it is committed time. Nothing puts it on the grid by default, so a planning session reads the gap as available and a block gets placed into a journey. This is not an occasional problem for anyone who leaves the building during the week, and it is the single most reliable source of a plan collapsing on the day. A calendar that measures the door to door journey and enters it as its own block makes the collision visible while the week is being planned rather than on the morning it fails, and the Travel time section of the guide covers how that measurement and its leave alert work.

The second is work with a deadline. An item on a task list competes with meetings for the same hours while being invisible to everyone, including the person who wrote it. Google Calendar now handles this directly: since November 2025, an empty slot can be turned into a block of time for a specific task, with its own visibility and do not disturb settings, and the task stays on the task list with reminders until it is completed. Google made it available to Workspace customers, Workspace Individual subscribers and personal Google accounts.

That feature is worth using for a plainer reason than productivity. A task with time on the calendar has a cost that can be seen. A task on a list has a cost that gets discovered.

Decide what refuses, and set it to refuse

A week plan is a set of intentions until something in the calendar is capable of saying no. Google Calendar has exactly two event types that do, and the difference between them is duration.

Focus time protects hours inside a working day. The Calendar API documents three modes for how it treats conflicts: decline nothing, decline all conflicting invitations including ones that already exist, or decline only new conflicting invitations that arrive while the focus time is present. It also carries a decline message, sent as the response when Calendar declines on the owner's behalf, and a chat status that can be set to available or do not disturb.

Out of office covers whole days away and declines with the same three modes and the same kind of message.

Entry	What it defends	Declines meetings
Ordinary event	Nothing, but shows as busy	No
Working hours	The edges of the day, as a signal only	No
Focus time	Hours inside a working day	Yes
Out of office	A whole day or more	Yes

Working hours belong in the table because they are so often mistaken for enforcement. Google describes the setting as a way to let people know the time is unavailable when they try to schedule, and they also feed the suggested times Calendar offers to other people. They refuse nothing. A week defended only by working hours is defended at the edges, and almost nothing that ruins a week arrives at the edges.

The mode to choose on a first attempt

Declining only new conflicts is the setting that produces information without producing damage. Existing commitments stay where they are, new ones bounce, and the pattern of what bounced is the first real data about how much of the week is genuinely negotiable. Declining everything conflicting reaches backwards into meetings that were already accepted, which is correct when the intention is to clear the block and destructive when the intention was to stop new bookings.

The booked half is more negotiable than it looks

Planning treats accepted meetings as fixed, and a share of them are not. Two mechanisms already in the calendar make that share visible without a single awkward conversation.

The first is the optional guest. Google Calendar lets an organizer mark a guest as optional from the guest list, and the marking exists precisely so that attendance can be a judgement rather than an obligation. Reading the coming week for meetings where the invitation itself says optional is a fast pass over the grid, and on a heavily invited calendar it frequently returns an hour or two. Doing the same as an organizer, marking the people who genuinely do not need to be there, hands the same hour back to somebody else.

The second is the response that was never given. Google Calendar has an account-level setting to be notified only about events answered Yes or Maybe, which means the unanswered invitations on a calendar are sitting in a third state: occupying the grid, producing no reminders, committing nobody. Those are the entries worth resolving during a planning session, because an unanswered invitation is a decision that has been deferred into the week rather than made before it.

Neither of these is about declining more. It is about the difference between a week that contains ten commitments and a week that contains seven commitments and three unresolved questions, which look identical on the grid and behave nothing alike.

Plan the week in the order the week will be lived

The sequence matters more than the method, and there are only four steps in it.

Look at what is committed, including travel and any task blocks from last week that were not finished. This is a reading step, not a decision step, and it should take two minutes on a calendar that shows a whole week at once.

Name what the week is for. One or two outcomes, not five. The count of usable blocks from the first section decides the number, which is why the counting comes first.

Place the blocks that defend those outcomes, with a decline mode chosen deliberately and a chat status set rather than left. The chat status matters more than the decline mode for most people, because the interruption that breaks a block usually arrives as a message rather than an invitation.

Leave one block unassigned. Every week produces something unplanned, and a week with no slack repays that arrival by eating whichever block was most important. The unassigned block is not padding. It is the reason the other blocks survive.

Repeating the plan without repeating the mistake

Weekly planning that works is the same review each week, which means the structure can be a repeating event rather than a decision to remember. Google Calendar handles that from the repeat control on any event, and the ceiling is high enough to ignore: repeating events have a limit of 730 occurrences, so a weekly planning slot runs for fourteen years.

Two habits keep the repeating version honest.

The first is to review before planning. The question that improves the next week is not what should be done, it is what happened to the last plan. Which blocks survived where they were placed, which were moved, and which were moved twice. A block moved three times has cost more attention in the moving than it would ever have delivered, and the honest action at the second move is to delete it and place it next week instead.

The second is to refuse to plan around a calendar that is not legible. Every judgement above depends on seeing a whole week at once: the full working day without scrolling, travel and tasks drawn alongside meetings, nothing hidden behind a click. When that reading takes a minute instead of ten seconds, the planning session quietly stops happening. The Features page covers how the views and the sidebar are arranged for that kind of reading.

The measurement that tells the truth on Friday

Planning advice is usually unfalsifiable, which is why so much of it survives without working. A week plan can be checked, and checking it takes less time than making it.

Three numbers are enough. Blocks placed. Blocks still sitting in their original slot on Friday. Blocks that ended up overlapped by something accepted later.

A survival rate below half means one of two specific things, and they have different fixes. Either the decline mode was too permissive, which is a setting, or the blocks were placed into gaps that did not really exist, which is a counting problem from the first step. A survival rate near one on a week that felt no different usually means the chat status was left available, so the hours were protected from meetings and not from anything else.

None of those three numbers is a judgement about discipline. They are a measurement of whether the calendar described the week accurately, and that is a fixable thing in a way that willpower is not. The reason the cost stays invisible without them is that a calendar records the meetings that happened and keeps no record of the blocks that were quietly moved out of existence.

What to change first

Count the usable blocks in next week before naming a single priority, and place one focus block with the mode that declines only new conflicts. Then look at it on Friday and count whether it moved. If the answer is that it moved because travel and tasks were never on the grid to begin with, a Mac calendar that puts them there automatically, such as Caltimate, is the thing to change next.