Time blocking: plan your day in blocks, then check the plan against what happened
Time blocking means deciding in advance what you will work on during each part of the day and writing it down as blocks on a calendar or a page. The plan will drift, so the part that makes it work is the review: compare the blocks with what you actually did, and change tomorrow's blocks by the size of the gap. You can do the comparison from memory, or against a recorded timeline, which is less flattering and more useful.
What time blocking is
Cal Newport describes the method in a post called Deep Habits: The Importance of Planning Every Minute of Your Work Day. His page divides each hour into two lines and fills the time with blocks labeled with specific assignments, with a right-hand column for notes and some extra room next to the blocks so he can correct the plan when the day unfolds in an unexpected way. He builds the schedule for the next day in ten to twenty minutes each evening, consulting his task lists, calendars and weekly and quarterly planning notes.
Timeboxing is a close relative. It allocates a maximum unit of time to an activity, a timebox, within which the planned work takes place, and it is used in agile project management and for personal time management. The difference in practice is the emphasis. A time block reserves a stretch of the day for a kind of work, while a timebox caps how long one activity may take. Most people who time-block end up using both.
Newport also notes that reactive work can be blocked like any other obligation. If your day is mostly responding to people, you can give open-ended blocks to that and still protect a small block, perhaps at the edge of the schedule, for deeper work.
Planning a day in blocks
Start from what is already fixed: meetings, appointments, the school run. Put those on the calendar first, then place the work that needs your best attention in the largest open stretches, and give the rest the smaller gaps. Label each block with the assignment, not the topic, so write draft the pricing section, not writing.
Plan the next day the evening before, as Newport does, and keep the planning itself short. Ten minutes is enough for a day with five or six blocks. Leave a margin beside the page or an empty row in the calendar for the correction you will almost certainly need.
If you have been recording your days, use that when you choose where hard work goes. Hina's Week tab has a Golden hours card that tells you which start hour accounts for the most deep work in the week, and when enough days have been recorded it words that as a suggestion to plan important work then.
Buffers and why plans drift
Plans drift for a documented reason. Buehler, Griffin and Ross studied the planning fallacy, a term Daniel Kahneman and Amos Tversky proposed in 1979, in a 1994 paper in the Journal of Personality and Social Psychology: people underestimate how long their own tasks will take, even when similar tasks took longer in the past, because they picture how the plan will go rather than recalling how past tasks went.
The same paper also tested a fix. In its fourth study the optimistic bias disappeared when people were asked to connect relevant past experiences to their prediction before making it, while people predicting someone else's task overestimated it and leaned on past cases. Your recorded days are exactly that kind of past experience, which is why the comparison below is worth more than a stricter plan.
A buffer is a response to that bias. One simple version is a 15-minute empty block after every two or three blocks, and an overflow block near the end of the day that absorbs what ran long. These are suggestions, not part of any official method. If the buffers are never used for a week, shrink them. If they are always used, your estimates are short.
The other reason is interruptions and the pull of other work, which a plan cannot predict. That is why the next step is the comparison.
Comparing the plan with the recorded day
Hina 0.6.0, a macOS menu bar app, records which app was in front, for how long, and counts of key presses and clicks, and stores them on your Mac. It takes no screenshots, asks only for the Accessibility permission, and keeps window titles and page paths off until you turn them on in Settings, Privacy. It does not read your calendar, so the comparison is something you do by looking at two lists side by side.
The Trail tab lists a day in hour groups. Each hour shows its active time, up to two main projects and, when you switched apps more than once, the number of app switches, and the rows underneath show each stretch with the time it ran until. A Show N away periods button reveals the idle stretches. The Day tab has Time by app and Deep work, which counts continuous stretches holding at least 10 minutes of work-category time. The day runs from 04:00 to 04:00 and the week starts on Monday, so work done after midnight and before 4 a.m. counts toward the previous day.
Put your block for ten o'clock beside the hour group for ten o'clock. Did the work you planned appear, in the app you expected, for roughly the time you gave it? If you plan an hour of writing and the hour shows eight minutes of a text editor and fifty of a browser and a chat app, the gap is information. The Daily report adds a Rhythm of the day section with the breaks between stretches and an Attention section with the most common switch, though earlier days and the Week tab are readable during the trial or with a licence, the report needs one even for today, and today's Trail and Day tab stay readable either way.
Recording itself keeps running offline. The app reaches the network in three places, each only when you click: opening the checkout page, activating a purchased key, and Check for Updates, which reads a version number from hina.day. None of them sends activity.
Adjusting tomorrow from the gap
Read each gap as one of three kinds. A block that ran long means the estimate was short, so lengthen that kind of block. A block that started late, or never started, means something else claimed the time, and the Trail rows show what. A block that was interrupted shows up as switches inside the hour, and the Week tab's Interruptions card names the apps, and the sites when page paths are turned on, where an interruption is a break under 5 minutes in the middle of work.
Then change one thing. Move the hard block to the hour with the most deep work. Turn the recurring interruption into its own block, which is Newport's own answer to reactive work: block it off like any other obligation. Or add a buffer after the block that kept overrunning.
Do not fix every gap. A plan that is tightened after every miss becomes a day you cannot keep, and the next miss is larger.
A one-week experiment
For five working days, plan three blocks each evening: one for hard work, one for shallow work and one for the thing you keep postponing, with a buffer after the first. At the end of each day open the Trail for that day, put it next to your blocks, and write one line: which block went to plan, which did not, and what took the time.
On Friday open the Week tab and read Deep work, Golden hours and Interruptions beside your five lines. Count how many of the fifteen blocks started within fifteen minutes of the plan and ended within a fifteen-minute margin, then pick the next week's single change. If Monday's plan was a fantasy, that is useful: you now know the size of the optimism you need to subtract.
What the record cannot tell you
The record shows apps and time, not intent. Fifty minutes in a browser might be research for the block or a rabbit hole, and a Trail row cannot say which. Work done away from the computer, such as a whiteboard session, a phone call or a meeting in person, appears as away. When Hina meets an idle stretch it cannot classify, it asks once on your return, with choices Meeting, Reading / thinking, Watching and Away, and your answer reclassifies that stretch.
So use the record for the questions it can answer, when work started, which apps held the time and how often you switched, and use your own notes for why. Time blocking is a way of deciding, and the recorded day is a check on the decision, not a score for you.