There's a fixed-price proposal in your inbox right now, or there will be soon. Most buyers think it caps what they'll spend. That's not what it does. It transfers the risk of getting the spec wrong from the buyer to the vendor, which is a much bigger deal than the cap, and the only reason fixed price is worth defending in an industry that quietly prefers to bill the hour. Here's what the contract is really doing under the hood, and how to tell whether the one on your desk is the real thing.
There's a comfortable fiction in software pricing: that time-and-materials is more flexible, more honest, and more accurate than fixed price, because it bills what was actually done. The fiction holds up only if you assume the buyer and the vendor have the same understanding of what needs doing, in which case the billing method doesn't matter much, because the bill will roughly converge. What time-and-materials actually does, in practice, is transfer the risk of mistaken scope from the vendor to the buyer. Fixed price transfers it the other way. That's the whole story; everything else is detail.
We use fixed price for almost all of our engagements that aren't pure advisory. Not because it's cheaper for the client on average, it isn't, slightly, but because the engagement is honest in a way T&M can't be. The price names what we think the work will take. If we're wrong, that's our problem to absorb. If we're right, the client paid exactly what we agreed. The cost of being wrong shows up on our balance sheet, not in a “discovery phase change order.” That's the discipline. Everything else flows from it.
What buyers think they're buying
Most procurement instincts treat fixed price like a budget cap. The number is the worst case; the work either comes in on time or the vendor eats the overrun. That framing is correct, but it's the smaller half of what's happening. The cap is, in dollars, a few percent of the value at stake. The bigger half is structural: what fixed price does to the conversation before the contract is signed.
A T&M proposal has a low scoping burden, because scope is the thing you're going to bill the hour against. The discovery phase becomes a profit centre. A fixed-price proposal has the opposite incentive, the vendor cannot afford to be wrong about what's being built, so they have to think harder about it before quoting. That thinking is value the buyer never sees on an invoice, but it is the most important value in the engagement. Vendors who quote fixed price have already spent money understanding your problem. Vendors who insist on T&M have not, because they don't need to.
So when a buyer chooses fixed price, they aren't choosing a cap. They're choosing a vendor who has to be right about scope before money changes hands. The asymmetry is the product.
What you're actually buying, the risk transfer
Every project has a baseline cost, the work that would happen in a frictionless world, plus a variance. The variance is what hits when something isn't what either side thought it was: a system that doesn't integrate the way the docs said, a workflow nobody mentioned in scoping, a regulatory clause that surfaces in week six. The variance is real, it's normal, and someone always pays for it.
In T&M, the buyer pays the variance, almost always, because the variance becomes additional hours and additional hours become additional invoices. The vendor's incentive is to keep finding the variance; the buyer's incentive is to wish it would stop. The relationship gradually adversarial.
In fixed price, the vendor pays the variance, by definition. So the vendor's incentive is to surface the variance before the contract is signed, to ask harder questions, to push for access to systems, to send a senior person to the early calls rather than a relationship manager. The conversation is allowed to be uncomfortable up front because the discomfort is cheap; finding the variance after the contract is expensive, and on the vendor's tab. The relationship is gradually cooperative.
AI tooling has tilted this further toward fixed price, not away from it. The variance cost on the vendor side is lower than it was five years ago: routine code, scaffolding, the long tail of straightforward integration work, all of that is now cheaper to absorb. So vendors can quote fixed price on bigger pieces of work without taking the kind of risk that used to make T&M the safe default. The buyers who notice get a better contract. The ones who don't are still being told fixed price is rigid and exotic. It isn't.
When fixed price is dishonest
Fixed price stops being honest in two situations, both of them about the vendor's incentives, not the contract's structure. The first is when the scope is genuinely undefined, the project is a research engagement, an exploration, a proof-of-concept where the answer might be “stop.” In those cases, a fixed price has to be priced so high to cover the uncertainty that the buyer overpays for the cases where the answer turned out to be easy. Better, for both sides, to be honest: a short discovery sprint, time-boxed and time-priced, with the fixed-price quote arriving at the end.
The second is when the vendor has cut scope quietly to make the price fit. The proposal says fixed price, the number is reassuring, but the deliverable inside has been silently narrowed. The buyer signs thinking they're getting the original ask; week six, they discover what they're getting is the cheaper version, and the “out of scope” conversations begin. This is a sales pattern, not a pricing pattern. The fix is to read the statement of work, line by line, and ask “what does this not include that we discussed?” at every section. A vendor who genuinely wants to deliver will answer it; a vendor who narrowed scope to win the work will start hedging.
There's a third edge case worth naming: the fixed-price-plus-change-order shape, where the price is fixed but the change-order machinery is so well-oiled that the final bill ends up looking like T&M anyway. The tell is the size of the change-order team relative to the build team. If the vendor has a department named after invoicing variance, you have a T&M contract wearing fixed-price clothing.
How to read a fixed-price proposal
Start with the price. Then read backward: what is the smallest scope that, when removed from the proposal, would force the price to drop. If that scope is well-described and clearly named, the proposal is honest, pieces can be cut, the price scales sensibly. If removing scope doesn't seem to move the number much, you're looking at padding. Either the vendor is unsure of their own estimate and built a buffer, or the proposal is bundled in a way that obscures the per-line value. Both are workable; neither is great.
Next, look at the assumptions section. A serious fixed-price proposal lists what it's assuming about your environment: which systems will be made available, which decisions will be made by which dates, which interfaces will exist and be documented. The longer that list, the more carefully the vendor thought about variance. If the assumptions section is two sentences, the vendor is going to discover the assumptions during the engagement and tell you about them as out-of-scope work.
Finally, look at the change-order language. The good version says “if scope changes, we'll re-price the affected milestones and discuss before proceeding.” The bad version says “change requests will be billed at the hourly rate listed in Appendix A.” The first is a partnership. The second is a T&M relationship with a fixed-price-shaped opening.
Fixed price isn't morally superior to T&M. Sometimes T&M is right, research, support, true ongoing operations. But when the engagement has a definable end and a definable deliverable, fixed price is the contract that forces both sides to mean what they're saying. That alone, before anyone talks about the number on the cover page, is what a fixed price actually protects.