Records are made before they are needed
Every contested final account arrives with worse records than either party remembers having. That is not carelessness. It is the predictable result of recording things when they seem important rather than when they happen.
The difficulty is that nobody knows in advance which week will matter. A delay that seems minor at the time becomes the pivot of a claim eighteen months later, and by then the people who were there have moved on and the detail has gone.
The answer is not to record everything, which nobody sustains. It is to record a small number of things consistently.
There is also a quieter benefit that has nothing to do with claims. A project where somebody is recording progress and resource weekly tends to run better, because the act of recording surfaces problems while they are still small. The claim value of the record is the second reason to keep it, not the first.
The five that carry weight
Instructions, with date, content and the basis of valuation stated. Progress against the programme, recorded at a fixed interval rather than when somebody remembers. Resource actually on site, which is what supports or defeats a disruption claim. Site conditions and events, including weather where it is relevant to the contract. And correspondence that responds to the substance of an issue rather than acknowledging it.
Each is a few minutes a week. Together they are the difference between an assessable claim and an argument about credibility.
The resource record is the one most often missing and the one most often decisive, because disruption is proved by comparing planned resource against actual, and neither number survives in memory.
Storage matters as much as capture. Records held in one person’s inbox, or on a phone that leaves with them, are records you do not have. A shared location with a simple naming convention costs nothing to set up and is the difference between having evidence and knowing that it once existed.
What a good record looks like
Contemporaneous, meaning made at the time rather than reconstructed. Specific, meaning it identifies the location and the activity rather than describing a general condition. And factual, meaning it records what happened rather than whose fault it was.
The last point is worth dwelling on. A record that assigns blame invites a dispute about the blame and distracts from the fact. A record that states the fact is far harder to argue with and considerably more useful later.
Photographs are useful and are weakened by not being dated and located. A photograph without either is evidence that something existed somewhere at some point.
The asymmetry nobody plans for
Contractors generally keep better records than employers, because their commercial teams are on site and recording is part of how they work. That is not a criticism of either party, it is a difference in structure.
The consequence appears when a claim arrives supported entirely by the other side’s documentation. Disproving a record you do not have an equivalent of is difficult, and the assessment tends to default towards the party who can evidence their position.
Correcting that costs very little. It requires somebody on the employer’s side to be maintaining an equivalent record, which is one of the quieter functions of a commercial appointment.
What to do when the records are thin
Something can usually still be proved. Payment applications, delivery records, subcontractor accounts, photographs held by different parties and the contractor’s own progress reports can be reconstructed into a narrative.
What cannot be recovered is precision, so the assessment moves from measurement to judgement, and judgement favours whoever documented more. We looked at what survives in thin records and construction claims.
The practical lesson is that the value of a record is highest for the events nobody expected to matter, which is exactly why the habit has to be routine rather than reactive.
What this means for you
Pick the five records above, decide who maintains each, and check once a month that they exist. That is the entire discipline.
If your project has been running for a year and nobody can produce a resource record or a dated instruction register, the position is weaker than the current reports suggest, and now is considerably cheaper than the final account for finding out.
Maintaining that record through the job is part of cost control and variations, and it is what makes the eventual final accounts and disputes straightforward rather than adversarial.
The test to apply is simple. If the person who has been on site every week left tomorrow, could somebody reconstruct what happened from what is written down. Where the answer is no, the project is carrying its commercial position in one person’s memory, and memory is the one form of evidence that cannot be produced later. Where we are appointed as client side quantity surveyor, maintaining that record is part of the role rather than an extra.