Both numbers were right.

What makes this difficult is not the gap. It is that both figures are about to leave the office, in different documents, to overlapping audiences, and somebody downstream is going to notice.

Why they diverge

The reasons are ordinary and they repeat.

Measurement rules differ. What counts as complete for progress is not always what counts as measurable for payment, and the second is usually stricter, because money depends on it.

Timing differs. The two close on different dates, which Week 14 covers from the commercial side. Work done near the boundary lands on one side in one document and the other side in the other.

Material at site is treated differently. Delivered material may carry value under the contract while representing no installed progress at all.

And the breakdown differs. The commercial structure maps to the contract; the physical one maps to where work happens. A figure that aggregates cleanly in one aggregates awkwardly in the other.

Checking before arguing

The mistake is to take the disagreement straight into a meeting. Almost every time, part of it dissolves against records that already exist.

The routine is the one from Week 8, applied at month end rather than weekly. Site quantities against delivery notes and store issues. Against the quantity surveyor's measurement. Against attendance, which caps what could plausibly have been done. Against store movements for the material actually consumed.

Usually the gap splits into three parts.

Some of it is a timing difference and will resolve itself next month. Some of it is a genuine measurement disagreement that needs a decision. And some of it is an error on one side, which is the part worth finding, because it is the only part that is still wrong after the meeting.

The material nobody agrees about

Of the four causes, material at site is the one that recurs most and argues hardest, so it is worth taking on its own.

Steel arrives and is stored. Under many contracts it carries value from the moment it is on site and accepted, so the valuation includes it. No progress has occurred: nothing is installed, nothing is measurable against a rule of credit, and the schedule has not moved.

Both positions are correct and they are not reconcilable by argument, because they are answering different questions. The only thing that resolves it is a decision taken once, early, about how delivered material is treated in each document — and then applied consistently rather than revisited every month when the gap reappears.

Who decides

Not the planner, and this matters more than it sounds.

When two records disagree, the item goes back to the person who reported it and to the department holding the contradicting record, in the same email, and it stays open until they agree. Project controls sets out what disagrees and by how much. It doesn't adjudicate.

There are two reasons for that. The obvious one is that the commercial position is a contractual matter and is not the planner's to settle. The less obvious one is that a figure quietly adjusted by project controls is a figure nobody else will stand behind. When it is challenged — by the client, by an auditor, by a claim two years later — the person who reported it will say it is not what they submitted, and they will be right.

The planner's authority here is not to decide the number. It is to refuse to publish two numbers that have not been reconciled.

Before, not after

All of this has to happen before either document goes out, which is an organisational constraint rather than an analytical one.

It means the two closing dates have to leave enough space between them for a short meeting, and that the meeting is about boundary items and differences rather than about the numbers in general. Twenty minutes, a list, and a decision on each line.

Do it afterwards and you are no longer reconciling. You are explaining why two published documents disagree, to people who now have a reason to distrust both.

When there is no time to reconcile

Sometimes the meeting doesn't happen. The dates were too close, somebody was away, the client moved the deadline.

The wrong response is to publish and hope nobody compares them. Somebody will, and usually it is the client, who receives both documents and has every reason to read them against each other. A difference discovered by the reader costs far more than the same difference declared by the writer, because the first raises a question about the numbers and the second only about the calendar.

So publish with the difference stated. One line: the two figures differ by this much, the cause is being confirmed, and the resolution follows next month. It is an uncomfortable sentence to write and it protects both documents.

System design

One table, filled in before either document is published, and only the lines that disagree go in it.

Record	Produced by	Required quality	Verified against	Feeds
Progress figure	Project controls	Built on rules of credit agreed at the start	Corroborating records	Report, schedule
Valuation	Commercial	Measured under the contract rules	Site measurement	Payment application
Difference	Both, jointly	Split into timing, measurement and error	Both records	The reconciliation
Agreed treatment	Commercial	Decided by the record holders, not by project controls	The decision	Both documents

Cause is the column that does the work. Timing resolves next month, measurement needs a rules decision, and error is the only category that is still wrong after the meeting ends.

Practical insight

Take last month's progress figure and last month's valuation and write the difference as one number. Most projects have never done this, because the two live in different documents owned by different people.

Then split it into the three parts: timing, measurement disagreement, and error. You won't manage it precisely and that is fine. The proportions are what matter.

If most of it is timing, you have a calendar problem and it is fixable by agreeing dates. If most of it is measurement, you have a rules problem and it needed settling at the start. If a meaningful share is error, you have found the thing that would otherwise have surfaced in an audit.

Key takeaways

• Progress and valuation answer different questions. Disagreement is the normal state, not a fault.
• They diverge for four ordinary reasons: measurement rules, closing dates, material at site, and breakdown structure.
• Check against records that already exist before taking the gap into a meeting. Much of it dissolves.
• Split the difference into timing, disagreement and error. Only the third one stays wrong afterwards.
• The planner sets out what disagrees and by how much. The planner doesn't adjudicate it.
• A figure quietly adjusted by project controls is a figure the reporter won't defend when it is challenged.
• Reconcile before publication. Afterwards it is not reconciliation, it is explaining two published documents that disagree.

Records born here. Month-end reconciliation sheet · boundary item list · the query email holding both parties.

What is coming next

Twenty-five weeks of inputs, outputs and the checks between them, and one question left that none of it answers.

Next week: the report nobody acts on — the most common failure in the trade, and what is worth changing when it happens.