Site records need the cause and the work held up, not just the fact of waiting

Every construction law update ends the same way: keep good records. It is the most repeated and least operationalised advice in the industry. Repeated, because it is genuinely how disputes are decided: in White Constructions Pty Ltd v PBS Holdings Pty Ltd[2019] NSWSC 1166 the claim failed for want of breach, and on delay, considered in the alternative, the court found that a well-kept site diary recorded waiting but not which work was held up (at [179], [205], [212]); the appeal was dismissed on liability without the delay findings being decided ([2020] NSWCA 277 at [131]–[135]); in Civil Mining & Construction Pty Ltd v Wiggins Island Coal Export Terminal Pty Ltd[2017] QSC 85, the contractor's own expert dropped his measured-mile analysis because the records did not support it (at [244]). And least operationalised, because “good records” is not a specification. This is the specification.

Tribunals give most weight to documents made when nobody knew there would be a dispute, and treat what is reconstructed afterwards with caution. A delay narrative assembled two years later from memory is advocacy; a diary entry made the day the crane stood down is evidence. The legal doctrines we write about elsewhere (common-sense causation, measured mile baselines, time-bar compliance, prolongation cost proof) all collapse to the same operational question: what did you write down at the time?

The discipline pays even when nothing goes wrong. Projects with real records settle claims faster and cheaper because there is less to argue about. The evidence pack is not dispute preparation; it is commercial infrastructure.

A diary that says “wet day, reduced works” loses claims. A diary worth keeping captures, daily, in a consistent structure:

• Labour: each crew or subcontractor, headcount, location, the activity actually worked, and where crews were idle or relocated, why, in plain words (“D&B gang stood down, awaiting RFI 114 response, moved to Level 3 fitout”).
• Plant: items on site, working or standing, and the reason for standing time.
• Progress: activities started, progressed (with measurable quantity where possible), completed.
• Events: instructions received (by whom, from whom, oral or written), access or information constraints, inspections, test results, incidents, visitors.
• Conditions: weather and its actual effect (“rain 10am to 2pm; concrete pour deferred; excavation continued”).

Two rules make it forensic-grade. First, the reason is the record: idle time without its cause proves nothing. Second, consistency beats eloquence: the same fields, every day, signed and dated, in a system that timestamps entries and prevents retrospective editing.

Around the diary sit five registers, each a one-page live document:

1. Notice register: every contractual notice issued and received: clause, trigger date, deadline, service method, response due. The defence against every time-bar trap we have written about, reviewed weekly.
2. Delay event register: each delay event as it emerges: cause, ownership (principal-risk, contractor-risk, neutral), affected activities, linked notices, status. This is what makes concurrency arguments winnable.
3. Variation register: directions (written and oral, with the confirmation request issued for oral ones), pricing status, approval status. After Kaloriziko Pty Ltd as trustee for Ryde Combined Unit Trust v Calibre Construction Group Pty Ltd (No 2) [2025] NSWCA 259, also record why each item is outside the contract scope: approving in-scope work as a variation does not make it one (at [118]).
4. RFI/information register: what was asked, when, answered when, and what work waited on it.
5. Daywork/resource records for disputed or changed work, signed daily where the contract allows.

The programme is statused honestly each month (actual starts, actual finishes, genuine percent-complete) and archived, every cycle, unedited. A contemporaneous record of what was driving completion each month is the raw material of every credible delay method; a baseline updated only when claims loom is an exhibit for the other side. Photos and video: dated, located, systematic (same vantage points weekly, plus event-driven capture), stored against the work breakdown, not scattered across personal phones that leave when the engineer does.

Cost records complete the pack: cost codes that can isolate time-related preliminaries by period and disruption by work front. Prolongation and measured-mile claims are built from exactly this granularity, and fail without it.

1. 01

   Recording the fact without the cause. A diary that says “wet day, reduced works” loses claims. Idle time without its cause proves nothing.

2. 02

   Leaving oral directions unconfirmed. The recurring entitlement-killer is the oral direction acted on and never confirmed.

3. 03

   Updating the baseline only when claims loom. A baseline updated only when claims loom is an exhibit for the other side.

4. 04

   Leaving photos on personal phones. Photos scattered across personal phones leave when the engineer does.

5. 05

   Editing after the event. The value of contemporaneous records is destroyed by any suggestion of after-the-fact editing.

• Specify it, assign it, audit it. A one-page records standard per project: the diary fields, the registers, the photo protocol, the programme cycle, with named owners and a monthly ten-minute audit. Systems fail at the level of ownership, not intention.
• Write reasons, not just facts. Train site staff that every idle resource, every resequencing, every deferred activity gets its cause recorded in plain English the same day. This single habit converts a diary from logbook to evidence.
• Capture the oral instructions. The recurring entitlement-killer is the oral direction acted on and never confirmed. Standing rule: every oral instruction is confirmed in writing within 24 hours, or sooner if the contract requires, and logged.
• Archive immutably. Monthly programme files, diary exports, photo sets: snapshotted and locked. The value of contemporaneous records is destroyed by any suggestion of after-the-fact editing.
• Run a quarterly “claim-readiness” pass. Pick one live issue and dry-run it: could we prove this delay path, this disruption, this variation entitlement from the records as they stand today? The gaps you find are cheap to fix now and unfixable later.

Get advice when you set up the records standard for a new project, when a delay, disruption or variation issue first emerges and you are not sure the records will prove it, and before you rely on site records in a claim. What has to be recorded, and by when, turns on the wording of your contract.