ingridleiria Weekly Review and Planning
Two serious commitments do not fit into one week by effort. They fit by deciding, in advance and in writing, which hours belong to which, and what will not happen at all. Without that decision the week does not become balanced; it becomes absorbed. The commitment with external deadlines and other people waiting takes every hour, and the commitment that only the person themselves is waiting on takes none.
That asymmetry is the whole problem, and it is invisible week by week. Nobody notices a Thursday when the thesis chapter did not get touched, because there was a genuine reason: a customer escalation, a board paper, a colleague off sick. The reason is real every single time. The cost only becomes visible at the scale of quarters, as a doctorate that has moved four weeks in nine months, a venture that never launched, or a person who has worked every evening for a year and cannot point to what the second commitment gained from it. By then the honest conversation is much harder, because the sunk cost has grown and the person is tired.
This is the review and the plan that make the decision explicit each week, on paper, so that the week is chosen rather than absorbed, and so that a bad pattern is visible in four weeks instead of four quarters.
A note on where this sits: it is filed in the writing track because that is where this library keeps the working methods I run on myself, but it is a personal operations skill rather than a writing one, and it is written for one person planning their own week rather than for a manager planning someone else's.
When to use this, and when not to
Use it weekly for anyone running two or more commitments that each have real obligations: a role plus a doctorate, a role plus a business, a role plus significant caring responsibilities, or two part-time roles. Use it when someone says they have too much on and does not know what to cut, when one commitment has been slipping for weeks, when two deadlines are converging, and monthly or quarterly in the extended form described below.
Do not use it to plan a team's work, allocate people across workstreams or track dependencies between them, which is program-management. Do not use it to design an organisation's meeting and reporting rhythm, which is operating-cadence-design. Do not use it to set quarterly objectives and measures, which is okr-planning; this skill takes those as given and decides which hours serve them. Do not use it to plan a single project end to end, which is strategic-plan-and-action-plan. Do not use it to decide whether to keep a commitment at all: that is a decision with options and a reversibility judgement, and it belongs in decision-memo, informed by the pattern this review produces.
What you need before starting
Last week's plan. The review is against what was planned, not against a general sense of the week. Without it the review becomes a list of what was memorable, which is systematically biased toward whatever went wrong most recently. Missing, for a first run: reconstruct last week from the calendar and sent messages in ten minutes, and note that the first review is weaker than the ones that follow.
The calendar for the coming two weeks. Fixed meetings, travel, teaching, anything with other people in it. This is what makes the hours arithmetic real. Missing: the plan cannot be made honestly. Get it before continuing, even if it means fifteen minutes of reconstruction.
Every deadline across all commitments for the next six to eight weeks, in one list. Merged, not one list per commitment. The collisions only become visible when a supervisor's deadline and a board date sit on the same page. Missing: build the list now; it is the single highest-value artefact this method produces and it is reused every week thereafter.
The open loops. Promises made to other people, decisions waiting on this person, unanswered messages that matter, each with the name and the date. Missing: spend ten minutes on inboxes and notes. Unretrieved open loops consume attention whether or not they are written down, which is why the list is worth the time.
The hours actually available. Working hours in the week minus fixed commitments minus the average unplanned load from recent weeks. Missing: for the first two weeks, use a placeholder of one fifth of working hours as unplanned load, and start recording the real figure in the review so the placeholder can be replaced.
One honest line on energy and health. The plan depends on it, and a person who has been running at the edge for a month cannot execute a plan built for a rested one. Missing: ask directly. This is a note for the person themselves, not a diagnosis, and where the line has been poor for several consecutive weeks the plan puts recovery before output.
The person's own good hours. When they think best, and which hours are reliably interruption-free. Missing: assume early morning before other people are working, which is the most commonly protectable window, and correct it after two weeks of evidence.
The method
The review comes first. Plans made without it repeat last week's mistakes with more optimism.
Part one: the review, 30 to 45 minutes at the end of the week.
List what got done, per commitment, as outcomes rather than effort. Not hours spent; the things that exist now and did not on Monday. A section drafted, a decision made, a hire closed, a dataset cleaned. Effort logs feel productive and hide the fact that a week can be full and produce nothing that persists.
List what was planned and did not happen, with the honest reason. Use a fixed set of reasons so the pattern is countable: no time, estimate was wrong, blocked by someone else, avoided, deliberately deprioritised. The distinction between avoided and no time is the important one and the one people most often blur. Over six weeks the counts say more than any single week's narrative.
Record what arrived unplanned and what it cost in hours. Escalations, requests, meetings added, a colleague's absence. Estimate to the nearest half hour. This number is the input to next week's arithmetic and is the reason most plans fail: the plan was made for a week with no interruptions, and there has never been one.
Update the merged deadline list. Anything moved, anything new, anything now closer than four weeks. Mark the ones that are externally imposed and cannot move, because those are the only ones with real priority claims later.
Update the open loops. What was closed, what is still waiting, what is now overdue and needs either an action or an apology. An overdue promise costs more the longer it stays unmentioned, and a message saying it will be late on Wednesday costs almost nothing.
Write the energy line, and read the last four. One sentence. Then look at the previous four weeks together. If the line has been poor for three or more of them, the following plan is built with a reduced hour count and includes rest as a scheduled item, not as whatever is left.
Part two: the plan, about 20 minutes at the start of the week.
Do the hours arithmetic before choosing any work. Working hours available, minus fixed calendar commitments, minus the unplanned load averaged over the last four weeks. What remains is the discretionary figure, and it is almost always between a third and a half of what people assume. Every later step is constrained by this number, which is the point of doing it first.
Name one outcome per commitment. The single result that would make the week a success for that commitment, written so that its completion is not a matter of opinion. Two commitments means two outcomes. Three outcomes for two commitments is the commonest way this method fails.
Schedule the deep-work blocks before anything else, starting with the commitment nobody is chasing. Two to three hours each, at the person's good hours, placed in the calendar as real entries. The rule for ordering is explicit: the commitment with no external deadline gets its block first and earliest in the week, because it is the one that will be sacrificed otherwise, and a block on Friday afternoon is a block that will not survive. Where a calendar tool is connected, create the entries so they defend themselves against other people's booking; where it is not, write them on the paper plan and decline conflicting invitations by hand.
Resolve collisions now, in writing. Compare the blocks against fixed commitments and move whichever loses, immediately, rather than discovering the conflict on the day. A block moved must be rescheduled in the same session; a block moved without a new time is a block deleted.
Read the merged deadline list for the next four weeks and pull work forward. Anything due in weeks two or three that needs a block now goes in now. This is where the method earns most of its value: two deadlines landing in the same week are survivable when they are seen three weeks out and fatal when they are seen on the Sunday before.
Build the shortlist, and price it. Five to eight tasks beyond the two outcomes, each with the day it will be done and an hour estimate. Add the estimates to the block hours. If the total exceeds the discretionary figure from step 7, cut the shortlist. The rule for cutting, in order: first anything nobody else is waiting on, then anything reversible if late, then anything that can be done badly and fixed later, and last anything with an external deadline. Sleep is not in the list of things that may be cut, and neither is the deep-work block, because cutting the block is how the pattern in the opening paragraphs begins.
Write down what will not get done. Explicitly, by name, on the plan. This is the step people skip and it is the one that makes the rest work, because an unwritten deferral keeps occupying attention as though it were still possible. Where someone else is expecting one of these things, tell them this week rather than next.
Name the people. Who needs to hear from this person this week: a supervisor, a manager, a collaborator, a family member. One line each on what they need to hear. A supervisor who receives one honest message in a bad week is easier to work with in the following one than a supervisor who receives silence and then an apology.
The hours arithmetic
The arithmetic is the part that keeps the plan honest, and it takes three minutes.
| Line | How to get it | Example | | Working hours intended this week | Decided by the person, not the calendar | 48 | | Fixed commitments | Sum of meetings, teaching, travel, appointments | 21 | | Unplanned load | Average of the last four weeks, from review step 3 | 6 | | Discretionary hours | Intended minus fixed minus unplanned | 21 | | Deep-work blocks | Two blocks of three hours | 6 | | Available for the shortlist | Discretionary minus blocks | 15 |
If the shortlist estimates exceed the last line, the shortlist is wrong, not the arithmetic. The commonest reaction is to plan the hours anyway and intend to work late; that decision should at least be made visibly, by raising the first line and writing down which evenings, so that the cost is on the page and countable at the next review.
Worked example
Situation. Rui Ferreira is head of operations at a 120-person software company and is four years into a part-time doctorate in economics. His supervisor had asked for a revised empirical chapter, about 9,000 words, by the end of the month. He had a board meeting on the 24th requiring a pack he writes himself, and a customer escalation had been running for eleven days. He had not opened the chapter file in nineteen days. His previous four weeks of reviews showed the thesis block cancelled in three of them, each time for a reason that was defensible on its own.
Task. Plan a week that would move the chapter, without the board pack being late and without the escalation getting worse. Good meant a measurable amount of chapter work completed, not a feeling of having tried.
Action. The arithmetic came first and was unwelcome. Intended hours 48, fixed commitments 24 including two escalation calls and a full day of interviews, unplanned load 7 as the four-week average. That left 17 discretionary hours. His initial shortlist, written before the arithmetic, came to about 31 hours of work, which is roughly the ratio the first honest arithmetic usually reveals.
The wrong turn was in the first version of the block placement. He scheduled the thesis block for Friday 14:00 to 17:00, which is where it had been for a year and where it had been cancelled three weeks running. The review's own reason column showed why: by Friday afternoon the week's unplanned load had already arrived and the block was the only movable thing left. The block was moved to Tuesday 07:00 to 10:00, before the office was active, and the board pack block was placed on Wednesday morning rather than the weekend.
The two outcomes were then written: for the doctorate, the results section rewritten with the new specification, about 2,500 words; for the role, the board pack complete and sent to the chief executive by Thursday evening for comment. Both are checkable by someone else.
The shortlist was cut from fourteen items to eight against the rule. Four items nobody was waiting on went to the not-this-week list. Two more were reversible if late and were moved to the following week, which put them on the same list. One, a hiring scorecard, could be done badly and improved later, so it was done in forty minutes rather than the three hours it had been given. The escalation stayed, because it had an external deadline and other people were blocked on it.
The not-this-week list had six named items on it, including a piece of analysis a colleague had asked for. Rui sent that colleague a message on the Monday saying it would come the following week, which took two minutes and removed the item from his attention entirely.
The deadline list, merged for eight weeks, surfaced the collision that mattered: the chapter deadline and the quarter-end reporting cycle both landed in the week of the 29th. Because it was visible on the 6th, three hours of chapter work were pulled forward into the current week and the following one, rather than being discovered as an impossibility three weeks later.
Result. The Tuesday block ran, and produced about 1,900 words of the 2,500 target. The board pack was sent on Thursday at 18:40. The escalation consumed 9 hours against the 7 that had been budgeted as general unplanned load, which is roughly what the buffer exists to absorb.
The chapter outcome was missed by a quarter. Rui's judgement was that this was still the best thesis week in two months, and the useful part was knowing exactly why it fell short: the 07:00 block produced good work at a rate of about 630 words an hour, which meant the outcome had been priced at four hours of work and given three. The estimate, not the discipline, had been wrong.
What did not resolve: the Thursday block, which had been intended as a second thesis session, was lost again. Over the following six weeks it was lost in four of them, and the pattern in the review log eventually forced a different conversation, about whether the role could be reshaped, rather than another attempt to protect a block that the week would not allow.
A second scenario, where it goes differently
The shape changes completely for someone with no employer. Ana Beltran is a freelance data analyst with four active clients, no salary, and a fifth commitment nobody pays her for: a training course she is building to sell, which is the only work not billed by the hour and the only work with nobody chasing it.
Two things about her week are different from the example above. There is no single calendar to subtract from, because four clients book her separately and none of them can see the others, so the fixed commitments have to be assembled by hand each Sunday and the unplanned load runs higher, at 9 hours in her four-week average rather than the 6 in the table. And the ordering rule in step 9 collides with money. The commitment nobody is chasing is also the one earning nothing this month, while a client deadline that moved forward by a week carried an invoice of 4,200 euros with it.
The arithmetic returned 12 discretionary hours against a shortlist of 26. The method does not settle the money question and should not pretend to. What it does is put the trade on the page. She took the client deadline, wrote one outcome for the week rather than two, and gave the course a single 90-minute block on Tuesday at 07:00 instead of the two three-hour blocks she had planned, on the rule that a week with no block is how a month with no block begins.
The declaration step works differently without an employer. There is no supervisor to write to, so the person she told was herself, in the review log, with a date: the course resumes in the week of the 20th. A commitment with no external observer needs its deferral written down and dated, or it is not a deferral at all, it just quietly stops, and that is the failure this whole file exists to catch.
What changed: with several clients and no salary the scarce resource is predictability rather than hours, and the rule that inverts is a different one. Ana cannot protect the unchased commitment first when a paying deadline sits inside the week, so she protects its minimum instead, one short block, and watches for repetition. Once is a hard week. Twice in a row is a scheduling problem to fix in the plan. Three times means the course is not a business but a hobby with a deadline attached, and that question belongs in decision-memo rather than in another weekly plan.
Output
Two artefacts, both short.
WEEK OF [date]
OUTCOMES
[Commitment A]: [the one result, checkable]
[Commitment B]: [the one result, checkable]
HOURS
Intended [n] Fixed [n] Unplanned load [n] Discretionary [n]
Blocks [n] Shortlist estimate [n] Fits: yes / no
DEEP WORK BLOCKS
| Day | Time | Commitment | Protected against |
SHORTLIST
| Task | Commitment | Day | Est. hours | Waiting on me? |
NOT THIS WEEK
[Named items. Who has been told, and when.]
DEADLINES, NEXT 6 TO 8 WEEKS, ALL COMMITMENTS
| Date | Commitment | What | Fixed or movable | Work pulled forward? |
PEOPLE TO CONTACT
| Who | What they need to hear | By when |
The review log is a single running document, one block per week, so the pattern is readable at a glance months later:
| Week | Outcomes met | Missed, and the reason code | Unplanned hours | Blocks held / planned | Energy line |
Failure modes
Planning without the arithmetic. Recognise it when the shortlist has no hour estimates. The plan will be roughly double what fits, and the overflow always comes out of the unprotected commitment. Do step 7 before step 8, every time.
The block that is always last in the week. Recognise it in the review log: the same block cancelled three weeks in four. Move it earlier and to a time before other people are active, rather than trying harder to defend it where it is.
Reasons that are all external. Recognise it when the reason column reads blocked or no time for six consecutive weeks with no instance of avoided. Avoidance is real and it hides inside legitimate reasons; the useful question is which single task was avoided and what about it is unpleasant, since that is usually a scoping problem in disguise.
The not-this-week list that is never written. Recognise it because the same five items reappear on the shortlist every week and are never done. Write them down as deferred and tell whoever is waiting.
Outcomes that cannot be checked. Recognise them by verbs like progress, work on, look at. Rewrite as a result someone else could verify.
Reviewing and planning in one sitting at the start of the week. Recognise it by a review that is generous about the previous week. The separation matters less than the order, but a review written on Monday morning is written by someone who wants to start, and it will not record the uncomfortable reason.
Buffer set from optimism. Recognise it when unplanned load in the plan is smaller than the average in the log. Use the measured average, not the intended one.
Treating the log as admin. Recognise it when the log has been kept for eight weeks and never read. The log exists to be read at the monthly review; unread, it is a cost with no return, and either the reading gets scheduled or the log gets simplified until it is worth reading.
Edge cases
A week with essentially no discretionary hours. Do not plan work into it. Plan the one thing that must happen, tell the people affected in advance, and protect a single short block. Then treat the next review as the place to ask why the week looked like that.
Illness, bereavement, or a family emergency. The method does not apply. Cancel the plan rather than reducing it, tell the people who need to know, and resume with a deliberately light week. A plan followed at half capacity produces guilt and little else.
Caring responsibilities that are unpredictable rather than scheduled. Raise the unplanned load figure to reflect them, place blocks at the most reliably protected hours even if those are not the best hours cognitively, and prefer two shorter blocks to one long one, because a three-hour block has three times the chance of being interrupted.
A travel or conference week. Discretionary hours are lower than they look, because travel fatigue is real and the evenings are not free. Plan for a third of a normal week's output and use the time for work that tolerates interruption, such as reading or review, rather than for the deep block.
One commitment is genuinely ending. For the final weeks, stop balancing. Give the ending commitment the hours it needs to finish cleanly, say so in the plan, and set a date when normal balance resumes.
Both commitments have the same immovable deadline. The plan cannot solve it. Decide which one is negotiated, negotiate it early with a specific new date, and take that decision to decision-memo if the consequences are serious.
The person is planning for someone else. This method is written for self-planning. Applied to someone else's week it becomes management, and the deep-work block loses its meaning if it was not chosen by the person who has to defend it.
Quality bar
- The hours arithmetic is on the page, and the planned hours are less than the discretionary hours.
- One outcome per commitment, each checkable by another person.
- Deep-work blocks are in the calendar before the shortlist exists, and the unchased commitment's block is the earliest.
- A written not-this-week list, with the people who have been told.
- Deadlines for all commitments appear in one merged list covering at least six weeks.
- The review was completed before the plan, and it records the unplanned hours and a reason code for every missed item.
- The energy line is written, and a run of poor lines has visibly changed the plan.
- The whole plan fits on one page.
Adapting this to your context
The numbers come from one salaried office-hours week, run by someone holding a full-time role and a part-time doctorate. The arithmetic transfers; the figures do not.
- The hours table. 48 intended, 21 fixed, 6 unplanned, 21 discretionary. Shift work, a teaching term or a clinical rota moves the intended figure weekly, so recompute rather than carry it forward.
- The unplanned load placeholder. One fifth of working hours for the first fortnight. Front-line and clinical roles routinely run at a third. Replace it with your own four-week average.
- The good hours assumption. Early morning before other people are working. Wrong for shift workers, for anyone whose colleagues sit in another time zone, and for people whose mornings belong to children.
- The number of commitments. Two, in both scenarios. Three will not fit, and the method's honest use then is to show which one to drop.
- The cadence. A 30 to 45 minute review and a 20 minute plan, weekly. Where deadlines move monthly, run it fortnightly and keep the merged list current between runs.
- What not to change. The order. Review before plan, arithmetic before choosing work, blocks before shortlist, and a written not-this-week list naming who has been told.
Related skills
program-management covers the same discipline for a team's work, with dependencies and owners, which this deliberately excludes. operating-cadence-design sets the organisational rhythm this personal plan has to survive. okr-planning supplies the quarterly objectives that decide which outcomes are worth naming. decision-memo is where a repeated impossible week goes when the real question is whether to keep both commitments in their current form. thesis-advisor sets the doctoral milestones this plan protects, and weekly-status-update is the outward-facing report to others, which is a different document with a different audience.