Never measure from the instruction. 3 timelines, a float test and a day count checked by hand. This is how a claim comes out at a fraction of the days claimed.
A delay claim arrives as a number of days measured from the instruction. That is the first mistake and it is the contractor’s, but it will become yours if you assess it the same way. The instruction date is not when the loss starts. I have had a claim come out at well under half the days claimed, with the whole difference in the day count. The only way it came out was by building 3 timelines and pricing them against the contractor’s own programme.
The contract sets a short period for the contractor to notify a delay once it knows, or ought to know, that an event will affect completion. Check the dates before anything else. If the notice was late, say so and reserve the position. But don’t pretend a late notice settles the claim if the delay was real and the record shows the Principal caused it. Reserving is honest. Hiding behind procedure when you know the answer is not.
Build 3 timelines. A is what the contractor’s own accepted programme had happening if the instruction had never been issued. Take it from the update nearest the instruction, using the activity IDs and their statused dates. It is the contractor’s own position, which is why it is hard to argue with. B is what should have happened once the instruction was issued, if the contractor had run the instructed work at a reasonable pace. Every step in B has to be sourced to something the contractor did or said. What it said mobilisation costs. What it actually achieved on the day it did the work. The designer’s actual turnaround. C is what actually happened, from the diaries, the subcontractor emails, the site camera and the correspondence.
Principal delay is B less A. Contractor delay is C less B. Both are counts of working days in sequence, not calendar windows. Say so, because on the calendar the contractor’s days usually overlap the days counted to the Principal. The contractor will take that point.
The count starts on the day the affected work would have started under A, not on the instruction date. The days between the instruction and that date were already occupied by work the contractor was doing anyway. Nothing was lost on them. Measuring from the instruction gives a number several days too high and hands the contractor the argument that you double counted.
A delay to an activity is not a delay to the contract. The question is whether completion moved. Set out the float on the affected path and at completion in the baseline and in the updates either side of the event, plus which chain was driving completion in each. Then answer the questions in order. Did the path have float before the event. Did it have float when the event happened. Would the instructed delay alone have moved completion at that time. Does it have float now. What moved between the updates. Whose days actually moved completion.
If the event only consumed float and forecast completion didn’t move, no extension is triggered at that time. Record it and monitor. If it exhausted the float and pushed completion past the due date, assess the portion beyond the float. Agree the float rules with the contractor early in the job, in writing, before there is a claim to argue about. On my current contract that is a 2 page memorandum both sides signed up to. It has settled every argument since.
Two delays running at once. The contractor’s later days being the ones that physically moved the forecast date doesn’t convert Principal delay already in place into contractor risk. Separate what each contributed rather than granting the period to whoever notified first. This is the hardest part and it is where the day by day table earns its keep over a summary.
A rain day is not a lost day without evidence that work stopped. Check what else went on. I have had the 2 heaviest rain days in a claimed window with piling continuing through both, drilling going ahead in rain on 3 other days and no day where the site stopped. No day was allowed. And where the contract carries a nil allowance for inclement weather, a genuine weather day is a weather matter, not an effect of the instruction.
Inclusive counting. Check every date range by hand, because 3 weeks on the calendar is not always 15 working days. A sensitivity that doesn’t add up, where moving timeline A changes the total but the contractor’s share is left at its original figure. Curing durations, where you have to decide whether day 1 is the day of the pour and then apply it identically in B and C, because getting it wrong moved a timeline by a full day. An update that isn’t statused, which is still usable as the no delay position but say so rather than glossing it. And quoting a meeting transcript when the issued minutes say something different.
Assemble the whole record before forming a view. Diaries for every day from the instruction to the affected work restarting. If the diaries folder is empty, that goes in the assessment rather than being worked around. Build the 3 timelines with every step in B justified in its own table. Test against completion in the 4 steps above. Then state the determination. Name the counter argument the other side will run and where the exposure is. The decision is the Engineer’s. The working has to be good enough that they can make it. What the ER can and can’t decide →
Send me the claim, the contract and the accepted programme as the native MS Project file, not a PDF. The logic matters more than the dates. Claim assessment →
Common questions
Because the days between the instruction and the date the affected work would have started anyway were occupied by other work. Nothing was lost on them. Measuring from the instruction overstates the delay and hands the contractor an easy argument that work it was doing anyway has been double counted.
A, no instruction, from the contractor’s own accepted programme. B, the instructed work handled at a reasonable pace, every step sourced to what the contractor did or said. C, what actually happened, from the record. Principal delay is B less A. Contractor delay is C less B.
Not on its own. Float in the accepted programme is there to be consumed by either party. The test is whether completion moved. If the event exhausted the float and pushed completion past the due date, the portion beyond the float is assessed.
Only if work actually stopped. Even then, check whether the contract already allocates weather risk elsewhere. Check what carried on that day. Piling continuing through the heaviest rain day in the window is not a lost day.
Separate what each contributed. Concurrent contractor delay does not reduce an extension otherwise due for Principal delay already in place. The contractor’s days being the ones that physically moved the date does not change whose delay came first.
Have a specific question
This page is general guidance. For a question about your own contract, use the free tool or get in touch directly.