Only procedure can be decided that early.

A subject the literature doesn't have

What happened when this week was researched is part of the argument. Across eighteen construction and contract references the kick-off appears about a dozen times, and the current project management standard mentions it once, inside a list of kinds of meeting. There is no guidance on the agenda, the attendance, the record, or what the meeting is for — while the same body of work names poor preconstruction preparation and poor communication among the leading causes of trouble on construction projects. The disease is documented. The one event held to prevent it isn't.

That gap is why this week argues from structure. What a meeting can settle is decided by when it is held, and that constraint can be reasoned about even where nobody has written the guidance down.

It also suggests why the guidance is missing. A kick-off isn't a technique, so it belongs to no discipline — not scheduling, not cost, not contract administration — and falls in the space between them, which is the space this track exists to cover.

What procedure actually means here

The decidable list is short and entirely about how the project will decide things later.

Where a notice goes, and to which named individual rather than which role. Who chairs which meeting, on what cycle. What a progress report contains and when it is due. Which transmittal route counts. What gets signed, by whom, before it is valid. How something moves up when two people at the working level disagree.

Every one is load-bearing for the life of the contract and reads, on the day, as housekeeping. That is the trap: the items with the longest reach sound least like decisions.

Take the first. A contract requires notice to the Engineer, and the Engineer is an office rather than a person. Whether notice reaches a named inbox, a shared one or a portal gets settled here, informally, in ninety seconds — and two years later a notice served correctly to the wrong one decides whether an entitlement exists.

Four ways an item leaves the room

Decided, with an owner and a date. Deferred to a named person by a date. Or deferred to further discussion.

The third never comes back — not because it was unimportant, but because the substantive items land there precisely because they matter and nobody yet has what they need to close them. Design completion by discipline, the access sequence, the interface with a package not yet let: all real, all unclosable this early.

What separates the second state from the third is a name. Reporting Week 23 takes apart what an action needs to survive. Why this meeting fills the third box is structural rather than carelessness: no other meeting combines full attendance with so little settled information.

And a fourth, which looks like the first

Those three are what the structure predicts. A planning engineer who has run this from the inside puts a fourth beside them, and it costs most.

His account: the reporting format, the progress measurement method, the responsibility matrix, the data submission dates, which data comes from which source — all discussed, some agreed. Then site carries on sending in the format site already uses. Planning keeps a different spreadsheet. The technical office keeps its own record. Equipment sits in a file of its own.

He sets out what has to stand behind a decision for it to survive, and it is a chain rather than a list: owner, format, source, deadline, validation, reporting. Leave any link out and the decision dissolves over time.

So an item can leave decided, with an owner and a date against it, and produce nothing. It appears in the minutes as a success and in the open list as closed, and decays anyway.

He puts it as the gap between taking a decision and turning it into something operable: a thing having been discussed at a kick-off doesn't mean it became a system.

What survives, and why

The same account answers what to do about it.

What lasts are the things with a clear owner and a daily or weekly rhythm. Daily reports prepared by particular people and sent to particular people keep running, because they are wired into the day rather than depending on anybody remembering a decision. What gets left to evaluate separately later disappears.

So the defence isn't a better minute. A decision that has to be remembered will be forgotten; one that has to happen before somebody can do their normal job won't. The useful question about anything agreed is which existing routine it attaches to — and if none, it needs one built for it or it won't hold.

Whoever called the meeting wrote the agenda

Nothing obliges an agenda to contain the items the other parties most need settled.

That isn't an accusation of bad faith. An agenda gets written from somebody's own list of open questions, and their list is genuinely different from yours — different obligations, different deadlines, a different view of what is unresolved. Items you need are absent because they were never on the paper the agenda came from.

The remedy costs an email: send your items in beforehand, in writing, as a request rather than an amendment. An item declined in advance is itself useful information, and cheaper now than in month eight, when discovering the question was never asked comes with whatever the answer would have prevented.

The minutes are the meeting

Six months on, nobody remembers the room; what exists is a document written by one party.

Nothing improper has to happen for this to matter. The chair records what they heard and what they think mattered, and two people can leave the same discussion with different and honest accounts of whether something was agreed or merely raised.

What matters is what happens next: minutes are issued, a period passes, nobody objects, and the account becomes the record. Reading them in the week they arrive is a small task; reading them in the week you need them is archaeology, and by then the version that stands is the one nobody challenged.

System design

Row two carries the whole difficulty of the week in its required quality. Minutes that say an item was discussed and minutes that say it was agreed describe the same twenty minutes and produce different projects.

Record	Produced by	Required quality	Verified against	Feeds
Agenda	Whoever called the meeting	Circulated early enough that others can add to it	The open questions of every party, not one	What the meeting can settle
Minutes	The chair	Distinguishes decided from raised, in the wording	What was actually said	The record, once unchallenged
Decision list	The same minutes	Each decision carries an owner and a date	The authority of the person named	How the project runs afterwards
Open list	You, within days	A name and a date against every item nobody closed	The minutes it was extracted from	When an item was first known to be open
Notice and escalation route	The meeting, or the contract	Named individuals, not roles, and a route when two levels disagree	The contract requirement	Every notice served for the next three years

Row four is the only one here nobody expects and the only one you own outright. It is also the second time this track has asked for the same instrument — week 8 wanted the basis of the programme, this week wants the open list, and both exist to put a date on something the project hasn't yet resolved.

Practical insight

When the minutes arrive, give them one pass with a single question in your head: what in here wasn't decided?

Pull every item that came out as deferred, noted, to be advised, or requiring further discussion. Against each one write the name of the person best placed to close it and the date it needs to close by. You aren't challenging the minutes and should say so in the covering line.

Then issue it, to your own team and to the chair, within a few days of the minutes reaching you. A list of open items is hard to object to, because on its face it is helpful — and it is.

Then do a second pass on the items that were decided, and ask a different question of each: what does the output look like, and who would notice if it stopped arriving? Where you can answer both, the decision will hold. Where you can't, you have found a fourth-state item, and it costs one line in an email now against whatever it costs when the report stops arriving.

What you now hold is a dated record of what your project knew it had not settled at the start. In six months, when one of those items has become a problem and the conversation turns to when it was first raised, your list is the answer — and it will be the only document anybody has that carries it. Twenty minutes, and the highest return available in this phase.

Key takeaways

• The kick-off is the one occasion the whole project is in a room together and it falls at the point of least knowledge, so almost nothing substantive is decidable and procedure is the whole of what is available.
• The literature documents poor preconstruction preparation as a leading cause of trouble and offers no guidance on the meeting held to prevent it.
• What is decidable is how the project will decide later: notice routes, chairs, cycles, report contents, transmittal routes, signature authority, escalation.
• Those items are load-bearing for years and sound like housekeeping on the day, which is the trap in the meeting.
• An item leaves in one of four states, and the third — deferred to further discussion — doesn't return.
• Substantive items land in that third state because they matter and can't yet be closed, not because anybody was careless.
• The fourth state is decided, owned and dated and still decaying, because without a format and a way of checking, everybody reverts to their own method within weeks — and the minutes record it as a success.
• What has to stand behind a decision is a chain rather than a list — owner, format, source, deadline, validation, reporting — and leaving a link out dissolves it.
• What survives is what has an owner and attaches to a daily or weekly rhythm, so it happens as part of the day rather than by anybody remembering.
• The question to ask of anything agreed is therefore which routine it attaches to, and if none, one has to be built: being discussed at a kick-off isn't the same as becoming a system.
• An agenda is written from one party’s list of open questions and yours is a different list, so sending your items in beforehand costs nothing and a refusal is itself information.
• Minutes issued and unchallenged become the record, so reading them in the week they arrive is a different task from reading them when you need them.

Records born here. The agenda and who set it · the attendance record · the minutes and the party who issues them · the decisions, each with an owner and a date · the open list, with a name against every item · the notice and escalation routes as actually agreed · the responsibility matrix, whether written down or assumed · for each procedural decision, its owner, format, source, deadline, validation and reporting route — and the routine it attaches to.

What is coming next

A format and a control mechanism for every procedural decision is a large ask for one meeting, and it isn't supposed to be the meeting's job. It is supposed to be written down already, in a document produced in the same weeks and opened by very few of the people it governs.

It states how the project will be run: who reports to whom, what the meeting structure is, how change is handled, how information moves. It can be assembled from the last project's version, approved without comment, and never opened again — which matters, because when two parties disagree about how something should have been done, it is the document that answers.