Fixed price or time and materials: what you are really choosing

A fixed price is not a discount, and it is not a gesture of confidence; it is a purchase of certainty, and the certainty has a cost that someone will carry. If the scope is written down and testable, buying certainty is rational. If the scope is a conversation, a fixed price does not remove the unknown — it moves the unknown somewhere you cannot see it. Paying by the day is not the cautious option either; it is the one that keeps the meter visible.

What is a fixed price actually selling you?

The sale is certainty about the total, from someone who accepts the risk of being wrong about it. The work is the same either way.

The vendor prices their own possible misjudgement into the number: a fixed price is a promise, and a promise made by someone who will lose money keeping it is a promise about to be renegotiated. That is arithmetic, not cynicism: you are paying for the possibility that the work takes longer than expected, whether or not it does.

The benefit is real: a total you can take to a bank without explaining a range. For a business that cannot absorb a growing bill, that is worth a premium.

What does an honest fixed price need to exist?

Three things, and without all three the number is a guess wearing a suit: a defined scope, an acceptance test, and a written change process.

A defined scope is a list of behaviours, not screens: what happens when a customer exceeds their credit limit, when a delivery arrives short. The moment a requirement says "manage" or "handle", the scope is open, and an open scope inside a fixed price is a disagreement waiting for a date.

An acceptance test turns "done" from an opinion into a check a stranger can run. For a website, published thresholds such as Core Web Vitals give the fixed price something measurable; for a warehouse process, it is a case from last month with real numbers.

A change process is the part buyers skip: who may request a change, who estimates it, who approves it, and what happens to the schedule when one is approved. Without it, every change is a negotiation conducted while the project is half-built and your leverage is lowest.

What happens to the unknown when it meets a fixed price?

It goes to one of three places: into the number, into an argument about scope, or into the quality of what is delivered.

Padding is the honest version: a margin for the unspecified parts, kept if the project goes well. You paid for a risk that did not materialise — that is insurance, not a scandal.

Scope games are the dishonest version, rarely planned. A requirement was vague, the vendor built the narrow reading, you expected the wide one, and the argument is now about what the document says rather than what your staff need. Both sides can be sincere and it still costs a month.

The quality drop is the quiet one. The system arrives on the date and is thin: no error handling, no test, documentation that lists the screens rather than the rules. It passes the acceptance test you wrote before you knew what to ask. This is the outcome that costs most, because it looks like success.

When is time and materials the safer choice for the buyer?

When the scope cannot be written honestly yet: exploratory work, a system nobody has modelled, a process the business is still arguing about.

A fixed price on unwritten scope does not protect you; it protects the vendor's interpretation. You pay for the worst case they could imagine and receive the narrowest version they can defend. Time and materials keeps the decision yours: you see what a week bought, and you can stop.

The real advantage of hourly work is the option to stop: after a month, after a prototype, after the first integration proves a bad idea. A fixed price removes it politely: cancelling means negotiating how much of the promise was already delivered.

It is also safer when you know you will change your mind: a fixed price cannot survive a buyer who discovers the requirement halfway through.

What do most real projects become?

A hybrid, and the honest ones say so at the start.

The usual shape: short discovery paid by the day, because its output is knowledge rather than software; a fixed price for the part discovery made definable; hourly work for change, with a ceiling and a review date. Some use a "not to exceed" figure, which transfers part of the risk and keeps part of the visibility.

What you are buying	Fixed price	Time and materials
The total	Known, if the scope holds	Unknown, but visible weekly
The unknown requirement	Priced in, or argued about later	Discovered, then decided
Your ability to stop	Costs a negotiation	Costs a conversation
What you must watch	The change process	The weekly output

The mixed shape is not a compromise for its own sake: it puts a fixed price only where something has been defined, which is the only place it means anything.

How do you decide, in order?

1. Write the scope as behaviours with their exceptions, before discussing money.
2. Try to write one acceptance test; if you cannot, do not fix a price.
3. Ask what the price assumes, and get the assumptions in writing.
4. Write the change process: who asks, who estimates, who approves, what slips.
5. Split the work: fixed price for what is defined, daily rate for what is not.
6. Put a stop condition and a review date in the hourly part.
7. Take the draft to a lawyer for the wording, before anyone signs.

Where this goes wrong

• A fixed price on an undefined scope. The vendor either loses money or delivers the narrow reading; the second is likelier.
• A change process nobody wrote. Every change becomes a favour, and favours add up to a slip nobody can point at.
• Acceptance tests that check features. "Can create an invoice" passes on a system whose totals are wrong.
• Choosing the lowest fixed price. The lowest number usually assumed the least, and the gap returns as change requests.
• Hourly work with no ceiling and no review. The meter runs, the scope drifts, nobody asks whether next week is worth buying.

What this does not cover

No figures: a rate is local, and a percentage is a guess dressed as a benchmark. It is not legal advice: the wording, the enforceability of a warranty, termination and ownership are for a lawyer, and this page only helps you ask in a useful order. Nor does it cover public procurement rules, which often prescribe the contract shape.

We build this way: a fixed price for what we can define and test, discovery and change by the day, and which is which stated before the contract is drafted.