Contract Terms That Decide Who Pays
A steel contract is usually assembled from three sources: a commercial template the supplier uses for every order, a technical annex written for this building, and a set of emails and drawings exchanged before signature. The first is long and generic. The third is short and specific. The argument, when it comes, is almost always about which of the three governs.
That is why the useful question is not whether the contract is fair. It is whether every place where the money can leave has a clause that names an owner, a trigger and a limit. Four such places exist on almost every steel order, and they behave differently enough to be worth separating.
Where a steel contract actually carries risk
A contract cannot make a bad building good. What it can do is decide, in advance and without emotion, who pays when a specific event happens. That requires the event to be describable. A clause that says the supplier will deliver in accordance with good engineering practice describes nothing, because two engineers can disagree about what that practice required and neither can be shown to be wrong.
Clauses are cheap before signature and expensive after
Every sentence added before signature costs a conversation. Every sentence added after a defect appears costs a negotiation with a schedule running against the buyer. This asymmetry is the whole reason to spend an afternoon on the document, and it is why a buyer who has signed four steel contracts tends to arrive with their own list rather than reading the supplier's.
The commercial template and the technical annex disagree by default
Commercial templates are written to protect the seller and are usually silent about measurement. Technical annexes are written to describe a building and are usually silent about money. Where the two meet, an order of precedence clause decides which wins. If the contract does not state that order, the answer defaults to whatever the local law says, which is rarely what either party assumed.
The scope definition is the clause that pays
Scope is not a list of what is included. It is a list of what is included plus a list of what is excluded, and the second list is the one that gets argued about. A contract that names only inclusions has defined half a scope, and the other half will be filled in later by whoever is holding the invoice.
Write the exclusions as tasks, not as nouns
An exclusion that reads "foundations not included" leaves open who supplies the anchor bolts, who supplies the setting templates, who checks the bolt positions before the concrete is poured, and who pays to re-drill when they do not line up. Each of those is a task with a cost. Writing them as tasks forces the buyer to see the work that has to be bought elsewhere, which is the point of the exercise.
The interface list is a scope document too
Every steel building meets other trades at defined points: the slab, the electrical containment, the fire protection, the process equipment. A contract that lists the interfaces and names the party responsible for each will settle most disputes before they arise, because most disputes are not about quality. They are about which trade was supposed to be finished first. A buyer who has not listed the interfaces has not bought a scope; they have bought a frame and a set of assumptions about who meets it.
Standard of work: the sentence that gets tested in year three
Standards are where contracts become quietly dangerous, because a reference to a standard is a reference to a document neither party has read in full. The practical approach is to contract to the properties that can be measured on delivery, and to name the standard only as the method of measurement rather than as the promise itself.
Contract to measurable properties, not to standard names
A clause that requires a coating to achieve a stated dry film thickness in microns can be checked with a gauge on arrival. A clause that requires a coating to comply with a named standard cannot be checked without laboratory work, and will therefore never be checked. The two clauses cost the same to write and produce very different after-sales conversations. Ask also which property will be measured, with what instrument and by whom, because a property with no method attached is a preference rather than a requirement.
Where the drawing and the standard conflict, say which wins
Drawings show a member size. Standards specify tolerances. When a delivered member is inside the tolerance and outside the drawn dimension, the contract decides whether it is acceptable. Naming the precedence costs one sentence, and it is the sentence that turns an argument about a millimetre into a decision about a document. Name the document, name the revision, and name what happens if a later revision arrives after fabrication has started.
Inspection and acceptance: the route, not the intention
Almost every steel contract mentions inspection. Very few describe a route, which means a sequence of hold points with the notice period, the party attending, the record produced and the consequence of a failed check. Without a route, inspection happens when someone remembers, and the record produced is a set of photographs that proves nothing in particular.
A hold point needs a notice period and a named party
A hold point that nobody attends is not a hold point; it is a delay. Writing "seven days' notice, attended by the buyer's nominated inspector, release issued in writing" converts an intention into an event. It also makes clear what happens when the inspector does not turn up, which is the case that generates the most ill will on a shipped order.
Define failure before you define remedy
A remedy clause describes what happens after a failure. It is only usable if the contract has already said what failure is: which measurement, against which document, with what tolerance, recorded by whom. Suppliers are not unreasonable about remedies. They are unreasonable about being told a building failed against a criterion that was never written down. The definition costs one paragraph and it converts every later argument into a measurement, which is the only form of argument that ends.
Limit of liability: read this one twice
The liability clause is usually the shortest clause in the document and the one with the largest financial effect. A common form limits the supplier's liability to the value of the contract, which sounds generous until the buyer works out what a delayed building costs them in lost production. Nothing in a steel contract is more worth a second reading.
Consequential loss is usually excluded, and it is usually the real loss
Direct loss is the cost of fixing the steel. Consequential loss is the cost of the building not being there, which on an industrial project is normally several times larger. A contract that excludes consequential loss has capped the buyer's recovery at the invoice, so the question to ask is not whether the exclusion exists but whether the price reflects it. The honest version of that clause states the cap openly and is priced against it, which is a commercial discussion the buyer can have before signature.
Retention and the defects period are the same clause in two halves
Retention holds money back until a defects period has run. The defects period only means something if the contract defines a defect, a notification route and a response time. A retention clause without those three is a payment delay with a legal name, and it will be argued about at the end of the project rather than at signature. Writing them together forces the response time to be decided at signature, when both parties still have something to trade.
Written by Johnny, who has read both sides of these clauses on orders that went well and orders that did not. If you have a contract in front of you and want the four clauses marked, send it over and I will return the places where the money is not yet allocated.
Related: payment milestones · red flags in a quotation · who does what between factory and site · what each line measures