GMP vs lump sum: choosing a construction contract when you build the same store fifty times

A lump sum contract fixes a single price for a defined scope, and the contractor carries the risk of building it for less. A cost-plus contract reimburses actual allowable costs plus a fee, and the owner carries that risk. A guaranteed maximum price is cost-plus with a ceiling, described by AIA as "a type of cost-plus agreement that establishes a price cap the owner will pay for the cost of the work, plus the contractor's fee," where the owner reimburses actual allowable costs "but only up to the guaranteed maximum price, unless the contract allows for adjustments" (AIA).

Most guidance then sorts these by project size and how complete the drawings are. That is reasonable advice for someone building one thing once. It is the wrong frame for a company opening its fiftieth store to the same prototype.

The three structures, in plain terms#

Unit price and time-and-materials arrangements exist alongside these and usually appear inside one of them, priced per unit for known repetitive work or hourly for undefined work. AIA's standard forms differ by delivery method too: A102 covers the traditional owner and contractor GMP arrangement, while A133 applies where a construction manager acts as constructor.

Contract selection is a legal and commercial decision with consequences that outlive the project, so run the structure and the specific language past counsel rather than choosing from a table, this one included.

The savings clause is worth more than the cap#

Under a GMP, the interesting question is not the ceiling. It is what happens underneath it.

AIA puts it plainly: "if the final project cost is less than the GMP, the contract should include a savings provision that explains how the remaining amount is distributed," and notes that some agreements return savings to the owner while others divide them on negotiated terms. That sentence contains the whole negotiation. A GMP with all savings to the owner and a GMP with a fifty-fifty split are different commercial deals wearing the same name.

Owners tend to negotiate the cap hard, because the cap is the number in the approval memo, and negotiate the savings split barely at all, because underruns feel hypothetical at signing. Across a programme they are not hypothetical. If a contractor is building your prototype for the twelfth time, an underrun is the expected case rather than a surprise, and the split determines who captures the benefit of your own repetition.

Two related terms deserve the same attention. What counts as cost of the work, since the line between reimbursable cost and overhead absorbed in the fee is where margin quietly moves. And how the contingency is treated at closeout, since a contingency that falls into savings is split under the savings clause, while a contingency that simply disappears is not.

Setting the GMP too early has a price, and it is contingency#

AIA gives a warning that is easy to read past. Contingencies should be established before construction begins, and setting the GMP too early "often increases contingency, which can raise the maximum price."

That is the mechanism behind a common and expensive pattern. An owner wants cost certainty at the earliest possible moment, so the GMP is set on a partial design. The contractor, pricing incomplete information, prices the uncertainty. The contingency inside the cap goes up, the cap goes up with it, and the owner has bought certainty about a number that is higher than it needed to be. The certainty is real. It just cost something, and the cost is invisible because it is embedded rather than itemized.

The same logic applies to lump sum. A fixed price on incomplete documents is not a fixed cost, it is a fixed price plus a change order stream, and the contractor has priced the gaps it can see.

There is no contract structure that removes the penalty for bidding unresolved scope. There is only knowing what the work should cost before you ask.

The question that actually decides it#

Here is the part the standard advice misses. The choice between these structures is usually framed as a question about the project. For a repeat-format occupier it is a question about the owner.

A contract type allocates risk between two parties who each hold a view of what the work will cost. Whoever holds the better view gets the better end of the allocation. On a one-off project the contractor almost always holds the better view: they have built more buildings than the owner has commissioned. On the fiftieth identical store the position reverses. The owner has fifty sets of actual delivered cost, by trade, by market, by vendor. The bidder has whatever it has built for you before, plus its own assumptions.

That changes the answer in both directions. With a strong independent cost basis, a lump sum stops being a leap of faith and becomes a straightforward test of whether a bid is above or below what the work has cost you before. Without one, a GMP is often the safer structure, precisely because it exposes actual cost and gives you a ceiling while you learn.

So the practical sequence runs backwards from how it is usually taught. Build the independent estimate first, then choose the structure that suits the confidence that estimate gives you. That is the argument behind building the budget before the bid set is complete, and the reason what your buildouts actually cost to deliver is a contract negotiation asset rather than a reporting exercise.

Contingency behaves differently in each structure#

Contingency is not one thing, and which pocket it sits in changes who benefits when it is not spent.

In a lump sum, the contractor's contingency is inside the price. You will never see it, you cannot audit it, and if the work goes smoothly the contractor keeps it. That is the deal, and it is not unreasonable, but it should be understood as a cost you paid rather than a reserve you hold.

In a GMP, contingency is usually visible inside the cap, which is the structure's main advantage. Unspent contingency flows into the underrun, and therefore into the savings clause, which is why the savings split and the contingency treatment have to be negotiated together rather than in separate conversations.

Separate from either is the owner's own contingency, held outside the contract for scope the owner adds. Keeping that distinct matters, because a programme that funds contractor-side risk and owner-side scope changes from the same pool cannot tell the difference between a contractor overrun and its own indecision. Distinguishing them is a scope and change order discipline, and the pattern across a programme shows up in change order management.

How to choose#

1. 01Establish the independent estimate first, from your own delivered cost history plus current market benchmarks, before any structure is discussed.
2. 02Judge the design's completeness honestly. If you would not bid it yourself, a contractor pricing it will add contingency, whichever structure you use.
3. 03Pick lump sum where documents are complete and your cost basis is strong, which for a mature prototype is most of the time.
4. 04Pick a GMP where real uncertainty remains and you want visibility into actual cost with a ceiling over it, then negotiate the savings clause as hard as the cap.
5. 05Reserve cost-plus for genuinely undefined work, and put a conversion mechanism in the contract so it becomes a GMP or a lump sum once scope is known.
6. 06Define cost of the work explicitly, including which overheads sit in the fee.
7. 07Level the bids against the same scope, not against each other, or you are comparing three different projects. This is the point of bid leveling and of bid evaluation.
8. 08Record the outcome against the estimate at closeout, by trade. That is what makes the next negotiation easier, and it compounds across a programme in a way no single project can show.

For tenant improvement work specifically, the structure interacts with the allowance and the landlord's approval rights, which adds a lease dimension to what otherwise looks like a pure construction decision. That overlap sits in construction management for tenant improvements, and the wider programme view in construction.

The short version: contract type is a risk allocation between two parties who each think they know what the job costs. Do the work that makes you the one who actually does.