The project record is written long before anyone needs it

When a capital project ends in a dispute, the outcome is determined substantially by documents created months or years earlier, by people who were not thinking about a dispute at the time.

That is the whole of the lesson, and it is uncomfortable because it means the decisive work happened when nobody thought it mattered. A party that administered the contract carefully arrives with a coherent record. A party that did not arrives with recollection, and recollection loses.

What contemporaneous means

The evidential weight of a document comes from having been created at the time, for an ordinary purpose, by someone with direct knowledge.

A daily record of labor on site, weather, and areas worked, written at the time, is powerful. A summary of the same period compiled two years later from timesheets and memory is not, even where it is accurate.

An instruction confirmed in writing on the day it was given is a fact. The same instruction recalled in a witness statement is a contested assertion.

A progress record showing what was complete at month end, agreed by both parties, settles a question that would otherwise be a matter of expert opinion.

This is why the record cannot be built when it is needed. Its value derives from when it was made.

The four records that decide most disputes

Projects generate an enormous volume of documentation and very little of it is evidentially useful. Four categories carry most of the weight.

Instructions and their confirmations. Every direction given, by whom, when, and what it changed. Verbal instructions confirmed in writing on the same day. Where an instruction is disputed, the party with the contemporaneous confirmation is describing a fact and the other is describing a memory.

Notices, and responses to them. Contracts contain notice provisions with time limits, and both parties routinely fail them. A notice given late may be invalid; a notice received and not answered may be admitted. Tracking both directions, with dates, is a small administrative task with a large consequence.

Progress and delay, recorded as they occur. Schedule updates with the actual dates achieved, the causes of variance stated at the time, and the effect on the critical path identified then rather than reconstructed later. Retrospective delay analysis is expensive, contestable, and largely unnecessary where contemporaneous updates exist.

Decisions and their basis. What was decided, by whom, on what information, and what was assumed. This is the record most often absent, and it is the one that distinguishes a decision that was reasonable on the information available from one that was negligent.

The record that wins a dispute was not created to win a dispute. It was created because somebody was administering the contract properly, and that is the only way it acquires its weight.

Why records fail

Rarely through concealment. Almost always through ordinary pressure.

Verbal instructions are accepted because confirming them takes time the project does not feel it has, and the relationship is currently good. Notices are not served because serving one feels adversarial while the parties are cooperating. Delay causes are not recorded at the time because the delay is expected to be recovered. Decisions are taken in meetings whose minutes record the outcome and not the basis.

Every one of those choices is reasonable in the moment and costly in aggregate. The project that maintains the record during the cooperative period is not being adversarial; it is being solvent about a risk it cannot predict.

The administrative habit

The discipline is short and it does not require a claims department.

Confirm instructions in writing the same day, in a consistent format, to the person entitled to receive them. Log notices in both directions with dates and response deadlines, and answer them within time even when the answer is a rejection. Record progress and the cause of any variance at each update, while the cause is known. Minute decisions with their basis, including what was assumed and what was still open.

None of that is claim preparation. It is what administering a contract consists of, and it happens to produce the record that decides the outcome if things go wrong.

The distinction matters for a practical reason. A project that starts keeping records when relations deteriorate has signaled its position and has a record that begins halfway through the story. A project that kept them throughout has a record that begins at the beginning — and, more often than not, does not need it, because a party with a complete record is in a much stronger position to settle early.