An extension of time claim has to establish three things: that a qualifying event happened, that it delayed completion rather than merely inconveniencing you, and that you notified it as the contract required. Every one of those is proved or lost on records made at the time.
The uncomfortable part is that most contractors hold plenty of records and still lose, because the records they hold answer the wrong question. Volume is not the problem. Specificity is.
Cause: what happened, and was it your risk?
The first job is proving the event occurred and falls within the relevant clause. Late information, a variation, exceptionally adverse weather, a statutory undertaker, an employer’s act of prevention: each has its own evidential shape.
What is needed is contemporaneous, specific and unglamorous. The date the information was requested and the date it arrived. The instruction and its timing. The actual weather conditions, recorded daily, against the works they stopped, not a general note that February was wet.
The weather example is instructive because it is the claim most often lost on records rather than on merit. To argue exceptionally adverse weather you generally need to show the conditions were exceptional against a baseline, and that they actually prevented work. That means daily records of conditions, of what was stopped, and of what the gang did instead. A diary that says “rain, poor progress” for eleven days proves nothing at all.
Effect: did it actually delay completion?
This is where most claims die. A delaying event that hits work with float in it may cost you money without moving the completion date, and an EOT compensates for the latter.
So you need to show the delayed activity was on, or drove the work onto, the critical path. That is a programme argument, and it needs three things most projects cannot produce:
- An accepted baseline programme, with logic, not just a bar chart issued once and never referred to again.
- Progress records good enough to show where the works actually were when the event struck. This is the piece the site diary provides and nothing else does.
- Updates showing how the delay propagated.
The second point is the one that turns on daily records. A delay analysis is only as good as the progress data underneath it, and progress data assembled retrospectively from memory and photographs is exactly what the other side will attack first. If you cannot show where the works were on the day, you cannot credibly show what the event did to them.
Timing: did you notify in time, and does it bar the claim?
This is contract-specific and the differences are larger than people assume.
Under JCT forms, the delay notice is generally a notification duty. Serving late is untidy and can affect what you recover, but it is not usually fatal to the entitlement in the standard unamended form.
Under NEC, the position is different. The early warning duty and the notification timescales for compensation events genuinely can bar a claim if missed, and that is the point of them.
Amended JCT is where people get caught out. Schedules of amendment frequently import a condition precedent that the standard form does not contain, converting a notification duty into a deadline that extinguishes the right. “We are on JCT so the notice is not critical” is an unsafe sentence unless someone has actually read the amendments.
Either way, the record that matters is proof of what you served and when. Not a copy of the letter: evidence of its transmission and receipt.
The records worth keeping, in priority order
| Record | What it proves | How it fails |
|---|---|---|
| Daily site diary | Where the works were, who was there, what stopped and why | Written weekly from memory; gaps in the disputed period |
| Progress photographs | Physical state on a date | Undated, unlocated, on a phone that has since been replaced |
| Notices and correspondence | That you complied with the contract machinery | Sent to the wrong person, or by a method the contract does not accept |
| Programme and updates | Criticality and propagation | No accepted baseline; updates never issued |
| Information request register | What you asked for and when it came | Kept in someone’s inbox, not as a register |
| Labour and plant returns | Resources actually deployed, and standing time | Cannot be reconciled with the diary |
The gaps matter more than the entries
This is the point that most surprises people. A diary kept diligently for nine months, with three weeks missing, is a liability, and the missing three weeks will be the ones around the delay event. Not because anyone did anything wrong, but because the busiest and most chaotic period on a project is exactly when the paperwork slips, and that period is also exactly when the dispute happens.
The other side does not have to prove your account is wrong. They only have to establish that your records are patchy enough to be unreliable, and then the tribunal is choosing between two reconstructions rather than reading yours.
Consistency across the whole job, including the boring weeks, is worth more than depth in the interesting ones.
What to change on Monday
Three things, none of which require new software to start.
Record conditions and consequences, not adjectives. “Heavy rain from 11:00, concrete pour to gridline 4 to 7 abandoned, gang of six redeployed to first floor blockwork” is evidence. “Bad weather, slow progress” is not.
Record where the works are, not just what happened. Progress by location, in terms someone off site could follow, is what makes a delay analysis possible at all.
Close the gaps. A short, complete, consistent entry every day beats a detailed one twice a week. If your process depends on the most stretched person on the project finding half an hour at seven in the evening, the gaps are structural and no amount of reminding will fix them.
That last point is the reason Construction Metric assembles the diary from the messages, photos and voice notes the team already sends rather than asking anyone to write one: how it works, or read a sample diary built from one ordinary day of site traffic.