Buyers choose fixed price to make a number certain, and are then surprised when certainty of price turns into uncertainty of outcome. The contract does not remove risk from the project; it decides who holds each kind and what everyone is incentivised to do when reality diverges from the plan. Understanding that trade properly is worth more than negotiating the rate.
01What each model actually transfers
There are several distinct risks in a software engagement, and no contract model transfers all of them. Fixed price transfers effort risk and nothing else. Time and materials transfers nothing by default but keeps everyone pointed at the same outcome.
| Risk | Fixed price | Time and materials | Capped T&M |
|---|---|---|---|
| Effort overrun | Vendor | Client | Vendor above cap |
| Wrong requirements | Client | Client | Client |
| Scope change | Client, via change request | Client, absorbed | Client, absorbed to cap |
| Quality shortfall | Client, usually | Shared | Shared |
| Schedule slip | Shared | Client | Shared |
| Vendor walking away | Client | Client | Client |
02What the fixed-price premium costs
A vendor pricing fixed work must cover the case where it takes longer than expected, and they price that uncertainty in. On well-specified work the loading is modest. On ambiguous work it is large, and it is invisible to you because it is baked into a single number.
Two vendors quoting the same scope at very different prices are usually pricing different levels of uncertainty, not different levels of value.
| Scope clarity | Typical loading | What the vendor is pricing |
|---|---|---|
| Precise spec, known systems, prior similar work | 10 - 20% | Normal estimation variance |
| Good spec, unfamiliar client systems | 20 - 35% | Integration and access unknowns |
| Outline spec, discovery not done | 35 - 60% | Requirement volatility |
| Legacy system, undocumented behaviour | 60 - 100%+ | Archaeology risk, or a refusal to bid |
03The incentives each model creates under pressure
Contract terms describe intentions. Incentives describe behaviour when the project is late, which is the state most projects spend some time in.
Under fixed price, once the vendor is behind their internal estimate, the cheapest route back to margin is narrower interpretation of scope and cheaper engineers on the work. Neither is bad faith - it is the structure operating as designed. Under time and materials, there is no structural pressure to finish, so the discipline has to come from your own governance rather than from the contract.
- Fixed price under pressure: scope disputes, quality reduction where it is not visible, staff substitution.
- T&M under pressure: comfortable velocity, expanding scope, no natural stopping point.
- Fixed price rewards writing the spec precisely; T&M rewards prioritising ruthlessly.
- Fixed price makes change expensive; T&M makes indecision expensive.
04Change requests are where fixed price is actually decided
On any fixed-price engagement of substance, 15-30% of the final spend arrives as change requests. That is normal, and it means the change-request terms matter as much as the headline number.
The number to interrogate is the change-request rate relative to the base rate implied by the fixed price. A vendor quoting $180,000 for work they have scoped at 900 hours has an implied rate of $200 an hour; if change requests are billed at $260, every change is a margin event and you should expect the scope boundary to be defended vigorously.
- Ask for the implied hourly rate behind the fixed price, and compare it to the change-request rate.
- Cap change-request rates at the implied base rate plus a stated percentage.
- Agree in writing who decides whether something is a change or a defect - and get an escalation path.
- Include a modest change allowance in the original contract, so small changes do not each become negotiations.
- Require change requests to be estimated within a fixed number of working days, or they block delivery.
05Capped time and materials, and how to make it honest
Capped T&M is the structure most sophisticated buyers converge on. You pay for actual effort, the vendor carries the overrun above an agreed ceiling, and you keep the ability to reprioritise without a contractual negotiation each time.
It only works with real transparency, because the failure mode is a vendor who quietly manages toward the cap and then presents an unfinished system with a straight face. The controls are simple and worth insisting on.
- Weekly burn against cap, reported without being asked.
- Named engineers with named rates, not blended anonymous capacity.
- A defined trigger - typically 70% of cap - at which scope is formally re-baselined.
- The right to reprioritise the remaining backlog at any point without a change request.
- A stated definition of done that includes tests, documentation and deployability.
06Choosing, in practice
Match the model to how much you already know, not to how much certainty you would like. Wanting a fixed number does not make the scope fixable.
| Work type | Recommended model | Why |
|---|---|---|
| Data migration with known endpoints | Fixed price | Genuinely specifiable |
| Integration against a documented API | Fixed price | Bounded and testable |
| Defined compliance implementation | Fixed price | The standard is the spec |
| New product build | Capped T&M | Requirements will move by design |
| Ongoing product engineering | T&M or dedicated team | No end state to fix |
| Legacy remediation | T&M after paid discovery | Effort is unknowable in advance |
| Proof of concept | Fixed price, small and time-boxed | The time box is the control |
Topics
Priya Iyer
Staff Engineer · SyncTrix
Writes about the engineering decisions behind production systems - architecture, delivery and the trade-offs that only show up at scale.
Building something like this?
SyncTrix engineers AI, SaaS, platform and cloud systems for enterprises and high-growth teams. Tell us what you're shipping and we'll scope it with you.