Variation Orders & Claims on GCC Projects: The Complete Guide

Change arrives as an instruction, or it arrives disguised

The clean case is easy: the engineer issues a variation order, the work changes, and the machine runs. That case is a minority of the changes on any real project.

The rest arrive disguised. A revised drawing that quietly adds scope. A verbal instruction on a Friday afternoon. A specification clarification that is in fact a specification change. A sequence the client requires that nobody has priced. Each of these is a variation in substance — the entitlement exists — and none of them has the one document that would establish it.

These are constructive variations, and the entitlement is real. Recovering it is a different matter, because the contractor must now prove that the work changed, that the change was directed by the employer or its agent, and that the cost and time consequences followed — all without the instruction that would have made it simple.

The discipline is unglamorous and it is the whole game: convert every one of them into a document, immediately. A confirmation of verbal instruction, sent within days, saying "you told me to do this, I am doing it, and unless you object within seven days it is on the record". It works because objecting is easy and silence is easy, but retrospectively denying an instruction you were told about in writing and did not dispute is very hard indeed.

A CVI raised three months later, when the work is already built, has lost the property that made it useful. The other side has no reason to accept it, and you have no reason to expect them to.

The notice is not a formality, and it is where good claims die

Under FIDIC 2017 the claim-notice window is short and it operates as a genuine condition precedent. Miss it and the entitlement can be extinguished — not weakened, not discounted. Extinguished, regardless of how right you were.

And the 2017 edition adds a requirement that catches competent people who have done everything else properly: a Notice must identify itself AS a Notice and cite the clause it is given under. An email that describes the delay event in complete detail, to exactly the right person, comfortably inside the window, is not a notice if it does not say that it is one.

The instinct that "they obviously knew" must count for something is overwhelming, and it is wrong. Actual knowledge does not cure a defective notice. The courts have been consistent about this for a very long time, and the reasoning is not perverse: the notice provision exists so that both parties know, definitively, that a clock has started and a claim is live. That certainty is the entire point, and it would be destroyed if informal awareness were a substitute.

Which produces the most expensive category of risk on any project: the one you correctly predicted, watched materialise, and then failed to notify.

Valuation: the hierarchy is the contract’s, not yours

Most standard forms set out an order of preference for valuing a variation, and it runs from most objective to least. Use the bill rates where the work is the same character under similar conditions. Derive a star rate from the bill rates where the work is similar but the conditions differ. Fall back to actual cost plus margin — dayworks — only where the work is genuinely unlike anything priced.

The hierarchy exists because each step down surrenders objectivity. A bill rate was competitively tendered. A star rate is argued. A daywork valuation is whatever the timesheets say, with no incentive to be efficient — which is exactly why employers resist it.

And the money moves at step two. The fight is almost never about whether the work is a variation; it is about whether it is "the same character under similar conditions". Pouring the same slab at night, in a confined area, after the tower crane has been demobilised, is not the same work — but the employer will argue that it is, because the bill rate is the cheaper answer.

Which makes the conditions a question of fact, decided by whoever recorded them. The contractor who documented the access constraint, the sequence disruption and the plant availability at the time has a case. The one who did not is arguing from memory against a rate that was competitively tendered, and will lose.

• Bill rates — same character, similar conditions. No argument available.
• Star rates — derived FROM the bill rates and adjusted. The derivation must be shown, not asserted.
• Dayworks — actual cost. Only where nothing comparable was priced, and only on signed sheets.

The daywork sheet is the entitlement

Cost-plus valuation rests entirely on daywork sheets: who was there, for how long, with what plant and what materials. And an unsigned daywork sheet submitted at the end of the month is an assertion, not a record.

This is the single most commonly lost entitlement in construction, and it is lost for the most mundane reason imaginable: nobody wanted to chase a signature on a Friday afternoon. The work was done. The cost was incurred. The money is simply not recoverable, because the one document required to recover it does not exist.

Get them signed at the time, by somebody with authority. Every time.

Time and money are two claims, and winning one is not winning the other

An extension of time protects you from liquidated damages for the extended period. It does not put a dirham in your account.

The money is a separate claim — prolongation cost — and it must be proved as actual cost incurred during the extended period: site establishment, supervision, plant on hire, preliminaries. Not as a percentage of anything, and not on a formula.

The recurring failure is exactly what you would expect: a contractor wins the extension of time, celebrates, and never makes the money claim — or makes it late, with no cost records tying spend to the extended period. The cost records had to exist at the time. They cannot be assembled afterwards, and a tribunal can tell when they have been.

Concurrency, and who owns the float

Where an employer-risk event and a contractor-risk event delay completion at the same time, the classic outcome — and the SCL Protocol’s position — is that the contractor gets time but not money. An extension, because an employer-risk event did delay completion. No prolongation cost, because the contractor would have been delayed anyway by its own concurrent culpable delay.

That is a default, not a law. Contracts increasingly draft around it explicitly, and jurisdictions differ, so read your clause before you build a case on the general principle.

And float belongs to the project, not to either party. It is consumed on a first-come basis — which has an uncomfortable consequence that few contractors internalise: a contractor that quietly spends float absorbing its own inefficiency has weakened its own position, because when an employer-risk event arrives and finds none left, the delay to completion is real and the argument about who caused it is the argument that ends up in arbitration.

The records decide everything, and they were decided two years ago

Every element above rests on the same foundation, and the foundation is a set of documents that had to exist at the time.

The delay analysis you can run is dictated by your records, not by which method is best. A project with a maintained programme, weekly installed quantities by location, and a diary that records what was NOT available can run a windows analysis or a measured mile — the two most persuasive methods available. A project without them will end up with an impacted as-planned analysis that the other side will take apart, because it models what should have happened rather than examining what did.

Which relocates the decision to where it actually belongs: two years earlier, on the day somebody decided whether to update the programme monthly and record quantities by zone. The claim you can make was chosen then, by people who had no idea they were choosing it.

• A diary that records quantities and constraints — not "progressed blockwork, 12 men, fine", which contains no fact a claim can be built from.
• Installed quantities by location, weekly. Without them, no measured mile.
• Hours allocated to activities. Unallocated hours inflate your burn and make your own crews look slower than they were — which understates the very claim you are trying to prove.
• Correspondence in a register, with notices served by the contract’s method to the contract’s addressee.
• Photographs with metadata, especially of work about to be concealed.

And the global claim, which is a symptom

A global claim attributes a total loss to a group of events without demonstrating causation between each event and each part of the loss. Tribunals dislike them and they frequently fail — and the objection is not procedural fussiness. If any material part of the loss was in fact the claimant’s own doing, the whole claim is contaminated, and there is no principled way to salvage part of it.

They are almost always a symptom rather than a strategy. A contractor makes a global claim because it did not keep records good enough to make a particularised one.

Which is the argument of this entire guide, compressed: the quality of your claim was fixed long before you decided to make it.