Fixed price vs time and materials is the wrong argument to have with a software supplier. A fixed price agreed before anyone understands the scope is a bet, and you are paying a premium for the supplier to take it or you are heading for a fight about change requests. Time and materials hands the whole risk back to you. The buy that works is scope first, then a fixed price for a build everyone can describe, with a change process that has prices on it.
We quote fixed. We also refuse to quote until the two weeks of discovery are done, and this post is why.
What is the difference between fixed price and time and materials?
A fixed price puts the delivery risk on the supplier, and time and materials puts it on you. Under a fixed price the scope, timeline and total are agreed before work starts; under time and materials you pay for hours at agreed rates.
Buying model | Who carries the risk | When it works |
|---|---|---|
Fixed price agreed before the scope is understood | The supplier on paper, you in practice, through a risk premium or a change-request dispute | Almost never for bespoke software |
Time and materials | You, entirely | Long-running product teams with strong internal ownership |
Discovery first, then a fixed price for the build | Shared: the supplier prices work it understands, you pay for changes you ask for | Most commissioned builds |
Most buyers ask for the first row, and most agencies agree to it.
Why does a fixed price agreed too early cost more?
Because work nobody has defined cannot be priced, so the supplier either pads the number or under-prices it and recovers the gap later. Steve McConnell's Cone of Uncertainty puts the error on an estimate made at initial concept at a factor of four in either direction, a sixteen-fold range from lowest to highest. A supplier who commits at that stage is carrying a two-to-four-times error and knows it. A cautious one prices the risk in. A hungry one wins the work and then finds that every clarification is a change.
The evidence that this shows up in outcomes is consistent. A 2017 peer-reviewed study in the International Journal of Project Management found the use of fixed-price contracts was connected with a higher risk of project failure than time and materials. [NEEDS SAM: the abstract wording was verified only from search snippets and the lead author's 2018 paper; open the DOI before publishing.] Boehm and Basili's landmark defect-reduction list (IEEE Computer, 2001) found projects spent 40 to 50 per cent of their effort on avoidable rework, and named "hastily specified requirements" as a leading cause.
The National Audit Office said the same about government in January 2025: pressure to deliver quickly means "contracts being awarded before the true requirement is fully understood", digital contracts are priced fixed "as if the supplier will provide a 'standard' service and take only limited risk", and "inflexible contracts can result in poor outcomes for both the department and the supplier". Swap "department" for your company and it describes the average agency engagement.
Is time and materials any better?
Only if you can carry the risk yourself, and most buyers cannot. Flyvbjerg and Budzier's study of 1,471 IT projects (HBR, 2011, a landmark) found an average cost overrun of 27 per cent, and one project in six overran by 200 per cent on average. Flyvbjerg's wider database of 16,000 large projects of every kind (2023) found 8.5 per cent delivered on time and on budget, and 0.5 per cent also delivered the benefits promised.
Time and materials without a defined scope is an open cheque against those odds. PMI's 2024 survey found scope creep on 30 per cent of projects on average, and failed projects lost 25.7 per cent of their budget. On time and materials, every one of those changes is your bill.
What does a fixed price need behind it to hold?
A scope that describes the work in enough detail that the number is a price rather than a guess. For a business system that means:
- Every system the new one has to talk to, and whether each has a usable interface, because integrations move the price more than features do.
- The actual data: the spreadsheets, forms and workarounds people use today, not the process as the org chart imagines it.
- Who uses it, by role.
- What is out of scope for phase one, in writing, and what phase two would cost.
- A change process with a price attached, so a new idea in week six is a decision rather than a dispute.
On a job-sheet system for a contractor, one of the outputs was a risk-assessment document that had to be generated for every job. Had it been variable content, it was a form builder, a different project at a different price. Discovery established it was a fixed template with a handful of merge fields, which is bounded work, and the fixed price held. Finding that in week one cost an hour. Finding it in month three would have cost the margin on the project, which is the point NASA made in 2004: a requirements error that costs one unit to fix when found early costs 21 to 78 units in testing and up to 1,500 in operation.
How long should discovery take before a fixed price?
Two weeks, in our case, and four to eight in the public sector. The GOV.UK Service Manual says four to eight weeks is typical and that stopping at the end of discovery "is not a failure". Our discovery is two weeks with the senior team, with no upfront cost, and it ends in a written recommendation you keep whether you build with us or not: what is worth building, what is not, how it phases, and roughly what each phase costs against our published ranges. Then we quote fixed, because by then we can.
It also produces the answer "do not build this". We talked a client out of a £100k build in exactly this fortnight. Flyvbjerg's rule for the projects that land is "plan slow, act fast". A fixed price is the "act fast" half. It only works after the other half.