Why weekly reviews die
The classic weekly review is a ceremony: clear every inbox, revisit every project, reflect in a journal, plan the horizon. As a quarterly exercise it is valuable. As a weekly obligation it competes with your Sunday evening - and loses, usually by week four.
The deeper problem is the data source. A review that asks "how did the week go?" is querying your memory, and memory is a genuinely unreliable witness to time. It compresses routine hours to nothing, inflates whatever was vivid or recent, and files the week under a single mood. Reviewing memory produces resolutions about feelings; only reviewing a record produces changes to a schedule.
The precondition: a record to review
An evidence-based review needs two artifacts from the week: what you planned, and what actually happened. If you time block, the plan already exists. The record can be as light as tracked start and end times - it does not need categories, tags or completeness. Even a record covering only your main working hours beats memory by a distance.
If you have neither yet, spend one week collecting them before your first review. A review without a record is a mood with an agenda.
Question 1: Where did the plan and the week disagree? (4 minutes)
Put the planned week next to the real one, day by day, and hunt for repeats - single misses are noise, patterns are signal. The same block starting late every day it appears. A kind of work that overran every single time. A block that exists in the plan and simply never happened all week - the ghost block, and it is trying to tell you something: wrong hour, wrong size, or wrong goal dressed up as a commitment.
Write each pattern down in one line. Not why - just what: "Deep work starts ~30 late." "Admin always overruns." "Friday workout: never happened."
Question 2: What ate the unplanned time? (3 minutes)
Now look at the hours that belonged to no plan at all. Some are life being life - noise, ignore them. What you are looking for is recurring unplanned time: the same person, the same kind of favour, the same afternoon leak appearing week after week.
Recurring unplanned time is a category your plan refuses to acknowledge. The fix is never resentment; it is a block. If helping your colleague reliably consumes three hours a week, a standing "help" block converts dark matter into schedule - same hours, but now they are chosen, bounded, and no longer detonating your afternoons at random.
Question 3: What one thing changes next week? (3 minutes)
One. The reviews that die are the ones that end with seven resolutions; seven changes cannot be attributed, and by Wednesday none of them are alive. Pick the pattern with the best ratio of annoyance to effort and make the smallest edit that addresses it: move the 9:00 block to 9:30, since 9:30 is when you demonstrably start. Double the admin block. Move Friday’s workout to the slot that survived every other day.
Note the change somewhere you will see it next Sunday. That is the whole output: one line of diagnosis, one edit to the schedule. Next week’s review opens by checking whether the edit worked - which is the moment the loop becomes compound interest.
Keeping it at ten minutes
Give the review itself a repeating block - ten minutes, same slot every week, typically Sunday evening or Friday shutdown. When the timer ends, stop; an unfinished review that happens every week is worth more than a thorough one that happens twice a quarter.
And hold the posture that makes the whole thing sustainable: the review is calibration, not judgment. A week where the plan and reality diverged badly is not a failed week - it is a well-documented one, and next week inherits the documentation.
- Ten minutes, timeboxed, same slot weekly - put a repeating block on it.
- Record first: no review without a plan and a track to compare.
- Patterns over incidents: only act on what repeated.
- One change per week, small enough to attribute.
- Open each review by checking last week’s change.
Key takeaways
- Reviews die from length and from running on memory; cap the time and review the record instead.
- Three questions: where did plan and reality disagree, what ate the unplanned time, what one thing changes.
- Recurring unplanned time is a missing category - give it a block instead of resenting it.
- End with exactly one small, attributable change to next week’s schedule.
- Check last week’s change first; that is what turns review into a compounding loop.
Doing this in Fezer
Fezer generates both review artifacts as a by-product of using it: the planned week exists because you block your days, and the record exists because tracking is one tap. Compare mode answers Question 1 visually - planned and actual, day by day on the same hours - and Analytics on the Week range rolls time up by front and goal, which is where recurring unplanned hours and starved life areas show themselves.
The review itself fits in a ten-minute repeating block, and everything involved stays on your device.