Contemporaneous evidence decides construction claims. Records made at the time carry weight; accounts constructed afterwards carry very little, in either direction.
That is a problem, because almost every contested account arrives with records that are thinner than the parties believe. The useful question is not whether the records are good but what can still be established from what exists.
Audit before you argue
The first exercise is an inventory. What documents exist, for what periods, in what form, and who created them. That includes the obvious sources and several that get overlooked.
Site diaries, minutes of progress meetings, instructions and confirmations, drawing registers, technical query logs, valuations and their supporting measures, correspondence, and the contractor's own cost records.
Then the less obvious: delivery notes, plant hire invoices, subcontractor applications, timesheets, photographs with metadata, and email threads that discuss the events without being about them.
Grade every head against the evidence
Each head of claim is graded on what supports it. Strong where the event, the notice and the effect are all documented. Partial where some elements exist. Unsupported where nothing contemporaneous survives.
That grading changes how effort is spent. Analysis goes into the items that can be proved. Items with no support come out of the claim or are accepted as negotiating make weights and treated as such.
This is the same discipline as testing entitlement, applied to evidence rather than to contract, and it belongs at the same point in the sequence. That order is set out in testing entitlement before you argue about value.
What can be reconstructed and what cannot
Physical facts can often be established after the event. What was built is measurable from the building. Quantities can be verified from as built drawings. Materials delivered can be traced through invoices.
Sequence can sometimes be reconstructed from dated photographs, delivery records and subcontractor accounts, even where no programme was maintained.
What cannot be reconstructed is what people knew and decided at the time. Whether an instruction was given, whether a delay was notified, whether a warning was ignored. Those depend on documents that either exist or do not.
Third party records
Where a party's own records are thin, other people's may not be. Subcontractor applications and correspondence, supplier delivery records, plant hire schedules and consultant reports all describe the same project.
These are frequently more reliable than the parties' own files, because they were created for an unrelated purpose and nobody was constructing a position when they wrote them.
Assembling them takes time and it is often the single most productive step in a thin record situation.
The honest conversation about value
A claim with weak evidence is worth less than the same claim with strong evidence, regardless of the underlying merits. That is uncomfortable and it is how disputes are actually resolved.
Telling a client that early changes the strategy: settle sooner, at a lower number, and spend less getting there. Equally, telling them late means the same settlement after two years of costs.
Where the position genuinely cannot be established, saying so is the most valuable output. That is set out under commercially contentious.
The lesson for the next project
Records cost almost nothing during a job and everything afterwards. A weekly record of labour allocation, area availability and instruction status takes a site team under an hour.
The same information reconstructed two years later takes weeks, produces a worse answer, and is worth less when it arrives.
That is the entire argument for record keeping as a commercial function rather than an administrative one, and the consequence of skipping it is described in variation claim exposure.
What this means for you
Audit the records before you form a position. Both parties usually discover their case is different from what they assumed, and the earlier that happens the cheaper it is.
Then chase third party records early, while people who held them are still findable. Our approach sits under final accounts and disputes.
Account in dispute and the records are patchy?
Send what exists. We will tell you what it supports before anybody commits to a position.