For about a year I did a weekly review every Friday and learned absolutely nothing from it.
I know it was every Friday because I have the notes. Forty-something entries, most of them variations on the same three lines: busy week, did not get to the writing, will protect Tuesday morning next week. Then the following Friday, busy week, did not get to the writing, will protect Tuesday morning next week.
I was not reviewing. I was journalling with a template. The pattern was sitting there in plain text, repeated over and over, and I never saw it because I never read the old entries. Each Friday I opened a fresh page.
A weekly review is only worth the time if it ends with something specific you will do differently, and the reliable way to get there is to compare what you intended against what your calendar and session history say actually happened. Recording the week is the easy part and does nothing on its own. Extracting one lesson from it is the whole job.
The practice has a legitimate origin. David Allen's Getting Things Done put the weekly review at the centre of the system, and the basic claim, that any personal system drifts out of trust without a regular pass over it, has held up well as an idea. What Allen describes is largely a maintenance routine, though, and maintenance is where most people stop.
Watch what happens on a Friday afternoon. You sit down to review, open your inbox because that is where the loose ends live, and forty minutes later you have replied to nine emails, filed some notes, moved four tasks to next week, and closed the laptop feeling productive.
None of that was a review. It was overdue admin wearing a review's clothes.
It happens every time because admin is concrete and reviewing is not. Filing a note has a visible result. Asking why the same project slipped for the third week running produces an uncomfortable answer and no tidy output. Given a free choice at 4pm on a Friday, almost everyone picks the inbox.
The fix is not to skip the admin. It genuinely needs doing, and trying to think clearly while eleven loose ends are pulling at you does not work well. The fix is to name the two halves and put a boundary between them.
Collecting is mechanical. Loose papers, open browser tabs, unfiled notes, things half promised in a chat thread, tasks that moved without being decided. You are sweeping, not thinking. Set fifteen minutes and stop when the timer stops, whether or not you are finished.
Reviewing is the part where you look at what happened and ask why. No inbox open. Nothing to file. Just last week's plan, last week's calendar, and a page for notes.
Doing them in that order helps because collecting clears the noise that would otherwise hijack your attention halfway through a thought. Doing them without a boundary means the review never starts, because collecting always expands to fill the slot.
You can run a useful review with three questions. More than that and the ritual bloats, which is the beginning of the end for it.
What did I actually finish? Not what did I work on, not what is in progress. Finished. This is deliberately harsh, because "made progress" is where weeks go to hide. A week with nothing finished is not necessarily a bad week, but it is a fact worth noticing rather than smoothing over.
Where did the estimate break? Find the thing that took the longest relative to what you expected, and be specific about the reason. Not "it took longer than I thought," which teaches you nothing. Something closer to: the draft took three sessions instead of one because I started without deciding the structure. That sentence is usable. The vague version is not.
What will I change next week? One change. Written as a concrete action, placed on a day. "Be more disciplined about mornings" is not a change, it is a mood. "Move the standing call from Tuesday 10 to Thursday 3" is a change.
If the third answer is empty, the review did not work. That is fine occasionally. Twice in a row means the questions have gone stale and you are answering on autopilot.
This is the single most informative step, and it takes about four minutes.
Open last week's calendar next to whatever you planned at the start of the week. The calendar is a record of what you really did, made by past you with no incentive to flatter present you. Memory of a week, by contrast, is heavily shaped by how the last day went and by how you feel right now.
Look for the mismatches:
Almost every useful finding I have had from a review came out of this comparison. The one I remember best: I had blocked deep work at 9am for a month, and the calendar showed something scheduled on top of it on eleven of twenty days. I had not been failing to focus. I had been failing to defend a slot, which is a completely different problem with a completely different fix.
Calendars tell you what was planned and often what was attended. They do not tell you how the work went, and that is where a review usually turns into guesswork.
This is the one place where a bit of instrumentation earns its keep. In Deep Focus I pull up the analytics page with the date range set to the last seven days and compare four things to what I remembered: how many sessions I actually completed, total focus time, the focus heatmap, and the distraction list.
The heatmap is the one that keeps surprising me. My memory of the week is a story, and the story is usually built around the two or three most emotionally loaded moments. The heatmap has no narrative, so it shows me that Wednesday was solid when I had written it off, and that the Thursday I remembered as productive was two short sessions and a long meeting.
Distraction tracking earns its place for a different reason. It records which apps kept surfacing during sessions, and that list is reliably different from what I would have guessed. I would have said social media. It was usually a messaging app and one specific news site, which is an actionable finding because both can go into a profile's block list in under a minute.
None of this replaces the thinking. It stops the thinking from being done on made up inputs.
Here is the thing that a single review cannot give you, and it is the reason the practice is worth keeping at all.
One review tells you about one week. Every week has a reason for going badly, and in isolation every reason sounds legitimate. Someone was off sick. The client moved the deadline. It was a heavy meeting week. Each of these is true and each explains nothing.
Six reviews tell you which explanations keep recurring. When "heavy meeting week" appears in four entries out of six, it is not an unusual week, it is your job's actual shape, and the correct response is structural rather than a promise to try harder. That finding is invisible from inside any single week and obvious from above.
So build the review to support reading backwards:
Long review notes have a hidden failure mode: nobody rereads them, including their author. A dense page written on Friday is a wall of text by the following Friday, so you skip it, the pattern stays invisible, and the review stops compounding.
Mine are four lines. Finished, broken estimate, repeat flag if any, one change. That fits on a phone screen and takes twenty seconds to scan.
Sitting alongside this, the forward-looking half of the week belongs in its own ritual. I keep it separate on purpose, because reviewing and planning interfere with each other, and I have written about the Sunday reset and how to plan a week that survives contact with reality elsewhere. Review is the input to that; it is not the same sitting.
I would rather say this plainly than pretend otherwise. Weekly reviews are dull. There is no completion sound, no visible output, and no immediate reward. Most people who start one stop within a month, and the usual explanation offered is a lack of discipline.
I do not think that is right. In every case I have watched, including my own, the review died because it was too long. Forty-five minutes of reflection is a big thing to schedule against a Friday afternoon when you are tired and the week is basically over. The first week you skip it, the entries stop being continuous, and once continuity is gone the whole value proposition collapses, because the value was always in reading backwards.
The fix is to shrink it until it is too small to skip. Fifteen minutes total: ten collecting, five reviewing. If even that dies, go to five minutes and three sentences. A five minute review that has run for eight months beats a thorough one that ran for three weeks, by a wide margin, because only one of them accumulates the history that makes patterns visible.
Two further honesties. A review will not fix a week that is genuinely overloaded; it will only describe the overload more accurately, which is useful but not the same as relief. And if your work is highly reactive, most weeks will look similar and the lessons will be thinner. In that case review monthly rather than weekly and spend the saved time elsewhere.
Fifteen minutes is enough for most people, split into about ten minutes of collecting and five of actual reviewing. The classic Getting Things Done version runs an hour or more, which is thorough and also the reason so many people abandon it. Start short. You can always extend a habit that is already running.
A review looks backwards at what happened and extracts a lesson. Planning looks forwards and commits to what will happen. They use different mental modes and blending them tends to produce guilt-driven overcommitment, so it is worth keeping them in separate sittings even when they fall on the same day.
Friday afternoon, while the week is still fresh and before the weekend blurs it. Monday morning is the common alternative and works, but you lose detail and you spend the freshest part of the week looking backwards. Avoid Sunday evening, where the combination of looking back and looking ahead tends to raise dread rather than clarity.
Three are enough: what did I actually finish, where did my estimate break and why, and what one thing will I change next week. Add a fourth only if it is checking whether any of this has now appeared two weeks running.
Usually because it ends with a general intention rather than a specific action with a day attached, and because old entries are never reread. Both are fixable in about three minutes: write one concrete change, and read the previous three entries before you write the new one.
The point of writing down last week is not to have a record of it.
It is so that in six weeks, when you read four entries saying the same thing, you finally believe it.
Get the latest productivity tips and Deep Focus updates delivered to your inbox
