Documenting delay correctly
A delay is rarely the real problem. It becomes one when it can no longer be evidenced in a dispute. This guide describes how to document delay so that it holds up later – and which mistakes regularly prevent that. It does not replace legal advice; assessing the individual case belongs with specialists.
1. Check the binding date
The starting point is not the delay but the planned state. A date is usually binding when it has been agreed contractually – as a completion date, as an expressly agreed interim date, or through an approved schedule programme forming part of the contract.
Without that reference point a delay can be observed but not asserted as default. This step therefore comes deliberately first: it decides whether the following ones carry any weight.
2. Document the planned-actual comparison
What is required is an objectively ascertainable deviation – a subjective impression on the part of site management is not enough. It shows, for example, in a clear lag in progress, in insufficient labour on site, in unexplained interruptions, or in following contractors being unable to start for want of preceding work.
The evidence is what is kept anyway: site minutes, photographic records, the site diary and the updated schedule programme. What matters is that the deviation is recorded with date and extent – not as an impression, but as the difference between two quantified states.
3. Examine and assess hindrances
Before any notice, it must be established whether the delay has a recognised cause. Weather, subsequent variation orders, late delivery of drawings, missing preliminary work by third parties, and notified and accepted hindrances usually come into consideration.
Where a justified hindrance exists, the contractor is not in default to that extent. The question of an extension of time is then settled beforehand, not afterwards. Leaving a notice of hindrance unanswered is the most expensive route: it does not go away, it resurfaces later as an accepted claim.
4. Allocate the shares of delay
Rarely does the whole delay rest with one party. It is therefore usual to divide it: the recognised share is expressly accepted as an extension of time, the remainder quantified and recorded as unexplained delay. This separation makes the position comprehensible – and open to challenge only where it should be.
5. Set an appropriate grace period
A grace period needs a concrete date and must be technically realistic. Too short, it misses its effect; without a date it remains indeterminate. It makes sense to require not merely a response, but a dated proposal setting out the measures by which the delay will be recovered.
The purpose remains restoring the agreed sequence, not escalation. A notice used as leverage rather than conducted as a finding loses precisely the objectivity it will later have to rest on.
Where it fails in practice
Strikingly, claims about time rarely fail for want of delay, but for want of documentation. Recurring patterns are:
- Notice without a binding date – without an agreed date the reference point is missing.
- Schedule programme never approved – a non-binding programme evidences no delay.
- Notice of hindrance left unhandled – the duty to examine was not exercised.
- Unclear or unrealistically short grace period – the intended effect does not arise.
- Mixing notices about time and about defects – two matters with different consequences in one letter.
- Notice given only verbally – without writing, nothing remains provable.
- Intervening too late – whoever watches for a long time will be asked why.
What links these points: they do not arise on the day of the notice, but in the months before. Anyone who never records the planned state bindingly and does not keep the actual state up to date can no longer produce either in hindsight.
This is exactly where a properly kept schedule comes in: versions with an approval date hold the planned state, the updated actual state holds the real course of events. In Menira every version remains as a state of its own – the planned-actual comparison does not have to be reconstructed later, it is already there.