The least informed week sets the reference.

What is in the folder, and what isn't

A planning engineer describes his first day on a nuclear construction site this way. What he opened first were the existing spreadsheets and reporting files, because there was no single system showing the whole of project controls. What he did next was ask people questions: who does the manpower figure come from, where does production come from, where does equipment come from, which file gets updated, which report goes to whom.

By the middle of the morning he had the finding that shaped everything after it. The data existed. The data flow didn't. Daily manpower in one place, an equipment list somewhere else, production quantities in a third spreadsheet; some held by individuals, some arriving by message, some obtainable only by asking.

By lunchtime the picture was that there was no working system to pick up and use. There was a system to be understood first.

And by the end of the day the question in his head had changed. It had started as how do I prepare this report. It finished as which data can I trust to prepare it, and where does that data come from.

That shift is the whole of day one, and it generalises past his job. A folder arrives. It holds the contract and its documents, a drawing set, a bill or a schedule of prices, tender correspondence, and a programme somebody produced for the bid.

Set against the six weeks before this one, notice what that list is. It is the conclusions of Phase A, and only the conclusions. The alternatives that were compared and rejected aren't there. Which figures were measured and which were adopted isn't marked. The estimate's accuracy class isn't stated. Which ambiguities the estimator had to resolve, and how, isn't recorded anywhere, because no document ever asked.

Which is why the day ends on provenance rather than method. Preparing a report is a skill, and the skill wasn't the obstacle — every figure needed for it existed somewhere on that site. What was missing was any statement of where each came from and which could be relied on without checking it against something else.

This is worth naming precisely, because the natural reading of a thin folder is that somebody did a poor handover. That reading leads to a conversation that goes nowhere. The material is missing because each stage passed its conclusion forward and kept its reasoning, and there was no field in any document for the rest. Nobody withheld anything.

The difference matters operationally. If the gap is a handover failure, the fix is to chase people. If it is structural, chasing produces apologies and no documents, and the time goes further reconstructing what can be reconstructed from what you already hold.

Two clocks, and they start on different days

The contract calendar is Contract Week 20's subject and it assembles the whole thing properly. One feature of it belongs here, because it shapes the first month rather than the contract.

The obligations of the opening weeks don't hang off a single date. Under the FIDIC forms the performance security is due within 28 days of receiving the Letter of Acceptance. The initial programme is due within 28 days of receiving the Engineer's notice of the Commencement Date. Those are the same duration measured from two different events, and the events can be weeks apart — the Commencement Date itself falls within 42 days of the Letter of Acceptance unless the contract says otherwise.

Two clocks, two anchors. A team that assumes one anchor discovers the mistake at the point where a deadline has already passed.

The dependency nobody draws

There is a link between two of those obligations that sits in different clauses and never appears on the same page.

The Employer gives right of access within the times stated in the Contract Data. And the Employer may withhold that right until the performance security has been received.

So the security isn't only a commercial instrument. It is, in the opening weeks, a gate on the site itself. A delay in issuing it — a bank taking longer than expected, a wording query, a group treasury approval — moves the date the works can start, and it does so through a mechanism nothing in the programme represents.

Due before the information exists

The initial programme has to be with the Engineer inside a month, and the Engineer may notice it back for not complying with the contract.

Consider what that programme needs and what exists at that point. Subcontractors unappointed. Vendor lead times that are enquiry responses rather than orders. Design status varying by discipline, which is what week 6 asked you to count. Access as a contractual date rather than a physical fact. Productivity inherited from a bid.

None of that is a reason not to produce the programme. It is a description of what the programme is: a structured set of assumptions, submitted as a plan, at the moment the fewest of those assumptions have been tested.

And the contract isn't asking for certainty. It asks for a programme within a period and gives the Engineer a mechanism to notice back non-compliance. Neither party is pretending the first issue is a forecast. The document simply survives past the point where anybody remembers that.

What you assume now is what you are measured against

Here is the consequence, and it is the whole week.

That first programme becomes the reference. Every delay analysis, every extension of time, every conversation about recovery for the next three years starts from it — and by then the assumptions inside it have hardened into something that looks like a considered plan rather than a set of week-one guesses that turned out to be load-bearing.

Which is the same defect this track has now met four times. The concept layout became a constraint. The feasibility assumption became site data. The estimate lost its accuracy class. Now a set of untested assumptions becomes a baseline — and each time, what was lost wasn't the content but the label saying how firm it was.

The difference is that this one is yours. It is the first instance in the whole sequence that you are in a position to prevent.

System design

Every row here is produced in the first month, which makes this the first table in the track you can fill in entirely yourself. Phase A’s rows all belonged to somebody else.

Record	Produced by	Required quality	Verified against	Feeds
Date register	You, in week one	Letter of Acceptance, Commencement, access, security deadline, on one line	The notices and Contract Data they came from	Every deadline in the first two months
Performance security	Your organisation, through a bank	Dated on delivery, because access may depend on it	The contract requirement	Right of access · the start on site
Initial programme	Planning, inside a month	Submitted as required, with what it rests on stated	The contract and the Engineer’s notice	Every later delay argument
Basis of the programme	The same submission	Plain sentences, dated, issued as part of the programme	What was known on the day	Your position when assumptions fail
Record of access given	Site, as it happens	Actual dates and extents against the dates in the Contract Data	What was physically handed over	Entitlement · the revised programme

Row four is the new one and it is the answer to the whole phase so far. Rows one to three exist on every project. The fourth is what turns the first three from a set of assertions into a set of assertions with a stated basis, which is the distinction Phase A lost at every stage.

Practical insight

Issue one extra page with your initial programme, and make it part of the submission rather than something you send alongside.

Title it the basis of the programme, and list in plain sentences what the programme assumes that nobody has confirmed: which subcontract packages are assumed and not let, which lead times came from enquiries rather than orders, which access dates are contractual rather than observed, which productivity rates were inherited from the tender, which design is assumed complete by a date. Ten or fifteen lines. Nothing rhetorical, no reservation of rights, no argument — just what the programme rests on.

It costs you an afternoon in a week where you have no afternoon, and you should spend it anyway, for two reasons.

It converts your least informed week from a liability into a dated record, which is the one thing Phase A never managed at any of its five stages. And it puts the Engineer on notice of your list while raising it is still an ordinary part of submitting a programme — rather than in month nine, when the same sentence reads as the opening of an argument.

Then get four dates onto one line where you can see them: Letter of Acceptance, Commencement Date, the access dates in the Contract Data, and the security deadline. That is the fixture list for your first two months, and it is four lines that take one afternoon to establish and are steadily harder to establish afterwards.

Key takeaways

• Influence peaks at the front end while the reference is fixed in the first month of delivery, so the moment of least knowledge produces the document of most authority.
• The folder holds the conclusions of the front end and none of the reasoning, and that is a property of the process rather than a poor handover.
• Day one ends on a different question from the one it started with: not how to prepare the report, but which data can be trusted and where it came from.
• The opening obligations hang off two different anchor dates — the Letter of Acceptance and the notice of the Commencement Date — which can be weeks apart.
• Right of access can be withheld until the performance security is received, so a banking delay moves the start through a mechanism no programme shows.
• The initial programme is due inside a month and can be noticed back for non-compliance.
• At that point subcontractors are unappointed, lead times are enquiries, access is a date rather than a fact, and productivity is inherited from a bid — so the initial programme is a structured set of assumptions submitted as a plan.
• It becomes the reference for every delay argument for years, by which time the assumptions read as a considered plan.
• This is the fourth time the track has met the same defect: what gets lost is the label saying how firm something was.
• It is also the first instance you are in a position to prevent, and one page attached to the programme prevents it.

Records born here. The date register, populated at last · the performance security and the date it was delivered · the initial programme · the basis of the programme · the record of access given, against the dates in the Contract Data · the priority of documents as actually established.

What is coming next

Somewhere in these weeks a room is booked and everybody is in it at once, for the only time on the project.

Things get decided there that will run for years, and other things get raised, noted as needing further discussion, and never returned to. What separates the two isn't importance. It is whether anybody left the room owning the item, and that is a property of how the meeting was recorded rather than of what was said in it.