The line about the night shift has now appeared in the open block of five consecutive updates. You wrote it on October 15. Ray hasn’t answered it. Every Thursday since, you’ve read it, thought that Ray still hasn’t answered it, and gone on to the next section.
So you remembered. Five times. What’s missing is the step after remembering, and a mailbox has only ever asked you to read.
By week seven a twelve week engagement has generated something like twenty of these. Things you said you’d do. Things you asked for. Things somebody agreed to out loud on a call, and nobody wrote down until you did, that evening, in a note. Most close quickly and by themselves, the way most things do. The handful that didn’t are still there. They’re sitting in messages you’ve already opened, in a condition your inbox has no way of showing you. The condition is open.
The Thursday update is an account of one week, and it’s good at being that. Holding a set of things across weeks is a different job, and no dated record does it. It’s a small job. It costs about fifteen minutes a month, once the items themselves are in a usable shape.
The failure is not the writing down
Almost everything written about action items is about capture, and capture is the cheap part. If an item made it into your note after the meeting, it exists. Everything on this page starts there. Getting it into the note is a craft of its own, and it lives back at the three next steps you settle before hanging up.
The expensive part is the middle: the four or five weeks between the day an item is made and the day it turns into a problem. The middle is invisible, for a reason worth understanding once. It’s why good people with tidy inboxes lose things anyway.
An email is a record of a moment. An action item is a state that lasts. Your mailbox keeps one piece of state per message, read or unread, and that’s a fact about you rather than about the item. Once you’ve read the message carrying the commitment, your mailbox considers the matter settled. Your mailbox answers one question, and answers it well: have you seen this. The item asks a different one.
So the attention curve runs backwards. An item is at its brightest on the day it’s made. It sits at the top of your inbox with a fresh timestamp and a person’s name on it. Five weeks later, on the day it actually matters, it’s nine hundred messages down, and the one thing that would surface it is you happening to think of it.
Where the items actually come from
In a delivery engagement the meeting is one of four sources, and it’s the only one anybody plans for. So a list built out of your meeting notes holds a little under half of what’s live.
The meeting or the call. The one you write up. This source has a habit attached to it, and that’s why it works.
Your own outgoing mail. Halfway through a reply about something else you write “happy to take a look at the reporting side while I’m in there.” That’s a commitment. It’s in writing now, on your client’s side, in a message you wrote and will never reread. They’re the ones holding it. This is the source that produces surprises, because it’s the one where the other party has the better record.
Their asks, buried mid message. The third paragraph of a message about badge access contains a request: check the aisle numbering against the 2025 drawings. That’s real work with a real dependency, and it arrived without a subject line of its own. What to do with those is a subject in itself. The short version is that the copying step is the one that fails when you’re busy.
The update you send. Every ask you put in the Thursday message is an item, and the update holds it for exactly one day.
Every week runs at least one of the four, and no single message holds them all. That’s the whole of why this wants a list rather than a habit of rereading.
The part that says when it ends
An item is a thing you can close. “Look at the picking data” will still be true in March, so that one’s a topic. “Send Dana the revised route map” stops being true the moment you hit send.
So every item carries a closing condition: the observable event after which it’s done. There are three.
You do something. The item closes when you deliver, send, or file the thing.
Somebody else does something. The item closes when a message arrives.
A date passes and the item stops being worth anything, whether or not anybody acted.
Knowing which of the three you have decides where the item belongs. An item that closes when you act belongs on your list of work, alongside everything else competing for a Tuesday. An item that closes when a message arrives does not belong there. Nothing you do on Tuesday moves it, and a list of work full of things you can’t work on is a list you learn to skim. Those belong on a standing request to be told when it lands, which is a quieter instrument.
The third kind is the one that slips past, and it’s the one costing you money right now. The night shift question was worth asking while the observation days were still running. Once they finished, it turned from a scoping question into a separate two day piece of work at a different price. Nothing announced that. The item sat there, looking the same as it had in week two, quietly worth less every Thursday.
Where an item has an expiry, put the expiry in the item. The expiry is the date after which you’d rather it were dead, and that’s a date of its own.
Every item carries two dates
The date it’s needed by, and the date you look at it. The email carries the first one. Treating them as the same date is the most common way a well written item goes quiet.
A due date is a fact about the world. The look date is a fact about your week. It’s the one that puts the item in front of a person while there’s still room to act on it.
For something you owe, set the look date a few days before the due date, far enough back that you can still do the work. For something you’re owed, set it a few days after the date they gave you. Chasing on the morning of a date the client named reads as a deadline you set for them rather than one they set for themselves. Three or four days later reads as a person keeping track.
For an internal decision inside their organization, where they gave you no date at all, the look date is whatever you decide it is. That decision is yours alone. Two weeks is usually right for a delivery engagement. It’s long enough to keep you clear of the supplier who writes every Monday, and short enough that a stalled decision surfaces while there’s still schedule to move.
None of this needs a system. Write the second date down in the same moment as the first, because that’s the moment it occurs to you.
What you owe and what you are owed
Two directions, one list, and they fail in opposite ways.
What you owe is the easy half to close and the expensive half to miss. It closes when you act, so it needs nobody’s permission. The reminder, when it comes, comes from the client, and by then the conversation has moved off the aisle map. It’s about whether you’re reliable.
What you’re owed is the loud half. You feel it every day, because you’re blocked by it. It’s also the half that working harder leaves where it is, which is why it accumulates. Five weeks of noticing produce exactly as much movement as one. The mechanics of that, and what brings a quiet thread back, are worked through elsewhere.
Keep both on one list. There’s only ever going to be one list you actually open. Read it in two passes, though, because the question changes. What can I close today. Then, what needs a nudge today. Ask both questions of every line and a twenty item list starts to feel like an hour of work, when it’s really about six minutes.
And expect the count to be skewed. Everyone overstates what they’re owed and understates what they owe. The reason is ordinary. The thing you asked for is vivid, because it’s in your way. The thing you promised was a subordinate clause you typed while thinking about something else. When your list says nine of theirs and two of yours, the two is wrong.
Three ways an item ends
Done is the one everybody models. There are two others, and they’re the reason lists rot.
Overtaken. The reason for the item disappeared. The exception log finally arrives, four weeks after you gave up and counted by hand. The item is dead. Nothing said so. It keeps appearing in your open block until somebody makes a decision about it.
Withdrawn. You decided against it, or they decided they’d rather not. This ending is fine, and common. It usually happens silently, and the silence is the part to fix.
A list that only has done in it fills up with items that ended the other two ways and still look live. Three or four dead lines is all it takes. After that you read the list loosely, because reading it carefully means rereading things you already resolved. That’s the real failure mode. The item is still there. What you’ve lost is a list you trust enough to open.
The expensive version is a withdrawal you keep to yourself. Your copy closes. Theirs stays open. A client with an item open on their side and closed on yours stays quiet about it until the closing meeting, and then it arrives as a complaint rather than a decision. So a withdrawal is a sentence you send.
Two things I’m taking off the list. I’m not going to check the aisle numbering against the 2025 drawings, because the observation days gave me better numbers than the drawings would have. And I’m dropping the night shift question for now: it’s moved past being a scoping decision, so if you do want it covered I’d rather price it properly than fold it in. Tell me if either of those is wrong.
Thirty seconds, in the next update, and both items are closed on both sides. That message is also the one place in this whole method where you get credit for the ones you dropped on purpose.
Reading the oldest ones first
The weekly reading catches this week’s items, which is what it’s for. Old ones slip past it, and that’s about how reading works rather than about you. An item you’ve read five times has stopped being readable. Familiarity does the work deletion would do, and it does it fastest on the lines that stay the same.
So the pass that matters runs in the opposite order. Once a month, and again before anything the client is going to read closely, open the list oldest first and stop at everything older than three weeks. There’ll be four or five. Fifteen minutes.
For each one, ask whether it should be on the list at all. That’s the question, ahead of what the status is. It has three answers, and all three are progress.
It’s live and it needs a nudge, which you send now rather than noting that you should.
It’s overtaken or withdrawn, so it comes off, with a line to the client if the client ever knew about it.
Or it’s genuinely stuck on somebody else, and a sixth week of repeating it in an update calls for a different move.
Do the pass before a milestone rather than after. In the week the client is already reading your interim findings, they’re paying attention, and an item raised alongside real work gets an answer. The same item raised in a quiet week in the middle of November gets skimmed by a person thinking about something else.
When a mailbox is the wrong place
Three cases, and all three want somewhere other than your own inbox.
More than one supplier owns items on the same project. Your list and the contractor’s list and the client’s internal list will disagree within two weeks, and the disagreement stays invisible to everyone. Ask which shared thing they already keep, and write into that, even if it’s worse than yours.
Several people on their side need to see the same status. A mailbox holds your copy. Mail four people a list of open items and you’ve made four copies that start drifting apart at once. The one they quote back to you in January will be whichever copy suits the argument. A client team that genuinely needs a shared view wants a shared surface.
The work carries an audit requirement. If somebody outside the engagement will one day need to see who agreed to what and when, that’s a record with rules attached, and email is a poor instrument for producing it on demand.
Outside those, for one consultant, one client, and twenty live items over twelve weeks, the mailbox is where the items already are. Everything else is a copying job that survives the easy weeks and fails in the week you needed it. That’s a different boundary from the one about schedules and dependencies, which the Thursday message draws where it belongs. A plan says what should happen. This page is about what stays open.
What Point keeps hold of
The judgment stays yours throughout. Whether an item still matters, whether to drop it, whether five weeks of silence means a problem or a busy client: that’s the work, and it belongs to you. What Point takes on is the holding, and the holding is the part that fails.
An ask that arrives inside somebody else’s paragraph becomes a dated item, and you copy it nowhere. It’s worth reading where the date comes from once, so the ones Point gets wrong are obvious to you. A commitment you made in your own outgoing message is held the same way. That matters more than it sounds, because that’s the source with no habit attached to it. The item you thought of yourself, in the car, you still type. What that costs decides whether it survives, and five seconds is roughly the threshold.
They gather in one place rather than one place per client, which matches how a consulting week actually runs. Each item sits with the conversation and the file it came out of. Opening it shows you the situation rather than a title, and in week nine that’s the difference between knowing what the aisle map item was for and guessing.
The other direction gets its own handling. Something you’re waiting on is tracked as waiting, and it comes back to you when the wait has run longer than it should have, rather than when you next remember. For a specific message you want to hear about the moment it lands, a standing request does it in a sentence. Between them, most of the items whose closing condition belongs to somebody else live somewhere other than your head.
Point leaves the question of whether an item is still worth having to you. The observation days closed the window on the night shift question, and Point has no way of knowing that. Point carries a dead item as patiently as a live one. That’s exactly why the monthly pass is a human fifteen minutes. Point wasn’t on the call either.
Point’s dial is set for each kind of action, starting in the middle position, where work is prepared and then held for you. The most valuable part of this is the noticing, and the noticing sits outside the dial altogether. An item arrives in front of you on the day you asked to see it, and that’s all that happens. Raise the dial on drafting a chase once you’ve watched a few and liked them. A message to a client about something they owe you is the kind of thing most people keep at review for a long time. That’s a sensible default, and it costs you nothing, because the saving is in the holding. Anything Point does on your behalf lands in a log with a timestamp against it. You can approve it, turn it down, or reverse it after the fact. One thing sits outside that, and every mail system shares it: once a message has reached your client’s server, it has reached it.
The flat list of what it does covers the rest, and the consulting shaped version covers the engagement rather than this one part of it.
The one that will not close
You do the pass on the Tuesday before the final report, oldest first, and it takes eleven minutes. Five items are older than three weeks. The aisle numbering one is overtaken and comes off. Two of Dana’s were done in October and never marked. The night shift question is dead in its original form, and it finally gets the sentence it should have had six weeks ago.
That leaves one, and it’s the newest of the five rather than the oldest. On November 12 you put the remediation phase in the update, priced, with a date on it. Nothing has come back. The night shift question sits in the same silence. Standing here at the end of the list, you can see that the two were one silence wearing two subject lines, and that you read it for five weeks as a scheduling problem.
Every instrument on this page is built for items that move when somebody is reminded. That one has been raised more clearly and more often than anything else in the engagement, so whatever is in its way, it’s something other than forgetting. A well held list has done its whole job here. It has told you, precisely and with dates, that this one needs a different instrument.