How to run a weekly commitment review

Most weekly reviews fail for the same reason most gym memberships do: they are designed by an optimistic version of you for a version of you that does not exist on Friday afternoon. Ninety minutes, twelve sections, a reflection prompt. It survives two weeks.

A commitment review is narrower and it survives, because it does one thing: it goes through every promise that is open in either direction and forces a decision on each. It takes twenty five minutes. Here is how to run it.

Set the constraints first

One fixed slot. Friday afternoon works because commitments made on Friday are the ones most likely to be dropped over the weekend. Monday morning works if your week starts with planning. What matters is that it is the same slot every week and it is in the calendar as a real appointment, not an intention.

A time box. Twenty five minutes, timer running. The box is not a productivity flourish. It is what stops the review turning into doing the work, which is the single most common way this fails. If you start writing the document you owe Priya, the review is over and eleven other items never got looked at.

One list, all sources. If your promises live in three apps, the review has to be over all three, in one place. A review that only covers Slack teaches you that Discord does not count.

The five passes

Work in fixed passes rather than item by item. Batching decisions of the same type is faster and produces more consistent judgments.

Pass one: capture what is missing, five minutes

Scan the places promises get made and add anything the list does not already have. You are looking for a specific verbal shape: "I will", "I can get you", "leave it with me", "let me check and come back to you", "send that over later". Also the questions you asked that never got an answer, which are the items you are owed.

Do not judge anything in this pass. Add it and move on.

Pass two: kill the dead ones, five minutes

Go down the list and delete. This is the most valuable pass and the one people skip.

An item qualifies for deletion when it no longer matters, when the person has solved it another way, or when it has been open for a month and you have declined to do it four weeks running. That fourth case is not failure. It is data. You have been making the same decision every week and simply not recording it. Recording it is a relief and it takes the item out of the way of the things that are real.

For anything you are dropping that somebody is still expecting, send one line now: "I am not going to get to this, do you still need it". That line takes eight seconds and prevents a much longer conversation later.

Pass three: date everything, five minutes

Every item that survives gets a date. Not a priority label, not a star. A date, meaning the day you will actually do it.

Items without dates are the ones that rot. A priority level tells you nothing about when, which means it tells you nothing about whether it will happen. "High priority, no date" is how something stays open for six weeks while feeling important the whole time.

If you cannot give an item a date, it belongs in pass two.

Pass four: chase what you are owed, five minutes

Go through the other direction of the ledger: what other people owe you.

For anything past due, send the nudge now, in the original thread, in one sentence with the original context attached. Not a new message in a new channel that requires the other person to reconstruct what you mean.

The reason this is a scheduled pass rather than an as-needed act is social. If you chase when you feel annoyed, you chase inconsistently and the tone carries. If you chase every Friday because it is Friday, it is procedure, and procedure is much easier for everyone to receive. Nobody minds a nudge that obviously happens to everybody.

Pass five: look at what closed, five minutes

Read the items that moved to done this week. Two questions:

How many of them were closed by a reminder rather than by memory? That ratio tells you how much of your reliability is currently held by a system rather than by luck. If it is high, the system is working. If it is zero, you do not have a system, you have a good week.

Which ones were late, and what was the pattern? Late items usually cluster. Everything from one client, or everything that required a decision you were avoiding, or everything you agreed to on a phone call. The pattern tells you what to change, and it is generally not "try harder". It is usually a capture problem at a specific moment in your week.

What good looks like after a month

You will know it is working when three things are true.

The list is shorter than it was, because pass two is doing its job. Reviews that only add are a growing pile, and a growing pile eventually gets abandoned.

You stop being surprised at deadlines. Not because nothing is ever late, but because you knew it was going to be late several days before it was.

And other people notice before you do. The clearest signal is somebody stopping their own follow-ups: they no longer ask whether you saw their message, because they have learned that things sent to you come back.

That is the entire return on twenty five minutes a week. Not more output. Fewer things that quietly did not happen.