Fixed price vs time and materials: which to choose for a software or AI project

Fixed price and time and materials differ in one thing above all: who carries the risk that the work takes longer than expected. With a fixed price, the supplier does, and you get a certain budget in return for defining the scope up front. With time and materials, you do, and you get flexibility in return. Choose fixed price when the outcome can be defined, and time and materials when it cannot.

This guide is for anyone buying software or AI development. It covers how each model works, where each one goes wrong, the hybrid that suits most projects, and how to compare quotes made on different models.

#The two models

Fixed price. You and the supplier agree a defined outcome and one price for it. If the work takes longer than the supplier planned, that is their cost, not yours. If you change what you want, the price changes through an agreed process.

Time and materials. You pay for the hours worked, at agreed rates, plus any costs such as software and hosting. There is an estimate, but the bill follows the actual time. You can change direction whenever you like, because you are paying for whatever gets done.

There are useful variants in between:

• Capped time and materials, also called not-to-exceed. Hours are billed as worked, up to a ceiling.
• Fixed price per phase. A larger project is cut into phases, each scoped and priced on its own.
• A monthly retainer. A fixed fee for a fixed amount of capacity, used for ongoing work.

#Side by side

#The incentives, honestly

Neither model makes a supplier good or bad. Each one does create a pull, and it helps to know which way.

Under a fixed price, the supplier earns more by finishing efficiently. The risk to you is corners being cut. You protect yourself with clear acceptance criteria, regular demos so you see the work as it develops, and a period after launch in which defects are fixed at no charge.

Under time and materials, the supplier earns more when the work takes longer. Most are honest, and the pull is still there. You protect yourself with a cap, a weekly report of hours against progress, and the right to stop at short notice.

A fixed price will usually include an allowance for risk. That is not padding. It is the price of someone else carrying the uncertainty, in the same way an insurance premium is.

#When fixed price fits

• You can describe what should be running at the end. "Customers can sign up, pay by card, create a project and invite colleagues" is a definable outcome.
• The technology is familiar to the supplier. They have built similar things before and can estimate from experience.
• The budget is a hard limit. For a startup spending its own money, or a department with an approved amount, certainty matters more than flexibility.
• You do not want to manage the work day to day.

Typical examples: a first version of a product, a single workflow automation, an AI agent with one well-defined job, an integration between two systems.

#When time and materials fits

• Nobody can define the outcome yet. Research, feasibility work, or "find out why this is slow".
• The work is inside an unfamiliar or older system, where the surprises cannot be estimated until someone has looked.
• The product is live and evolving. Priorities change every week based on what users do, and a product owner on your side decides what to build next.
• You have the time and the ability to steer. Time and materials rewards an involved, decisive client, and punishes an absent one.

#The hybrid most projects should use

You rarely have to choose one model for a whole engagement. A sequence that works well:

1. A small, fixed-price discovery. One or two weeks to turn an idea into a defined scope: what will be built, what will not, the risks, and a plan. This is cheap, and it removes most of the uncertainty that makes fixed pricing risky for either side. You should own the output and be free to take it to any supplier.
2. A fixed price for the build, or for each phase of it, based on that scope.
3. A monthly arrangement for ongoing work after launch, when the work becomes a stream of small improvements that nobody could scope in advance.

Each stage uses the model that fits the amount of uncertainty at that point.

#How change works under a fixed price

The fear with fixed price is being locked in. In practice a good change process has three routes.

• Swap. A new requirement replaces something of similar size. The price does not change. This covers most changes in a well-run project.
• Add. The change is estimated and priced, and you approve it before any work is done.
• Park. The idea goes on a list for the next phase. Many parked ideas turn out not to matter once real users arrive.

What makes this work is a scope document precise enough that both sides can tell whether something is a change. That document, not the pricing model, is what prevents disputes.

#What a good fixed-price agreement contains

• The outcome, in a sentence or two.
• The scope, as a list of what users can do or what the system does.
• The exclusions. What is not included is as useful as what is.
• Acceptance criteria. How you will both know each item is done.
• Assumptions, such as "the client provides API access to the CRM in week one".
• The change process.
• The timeline, with regular demos, not only a final delivery date.
• Ownership. The code, the accounts and the data are yours.
• What happens after launch. A period in which defects are fixed at no charge, and how long it lasts.
• Third-party costs, such as hosting and AI model usage, stated or estimated.

For a time and materials agreement, replace the scope and acceptance criteria with rates, a cap, a reporting rhythm and a notice period.

#Comparing quotes made on different models

A fixed quote of $30,000 and a time and materials estimate of "about $22,000" are not the same kind of number. To compare them:

1. Ask the time and materials supplier for a range and a cap. "About $22,000" often turns into "$22,000 to $35,000, capped at $38,000", and that is the honest comparison.
2. Check that the scope is the same. Does each one include testing, deployment, documentation, handover and a period of fixes after launch?
3. Compare the cost of the outcome, not the hourly rate. A higher rate from a faster team is often cheaper.
4. Ask each supplier what would make the project cost more. The quality of that answer tells you a lot.

#How we work

So that you have a concrete example: every SolvSpot project is quoted as one fixed price before work begins, and every build includes a 30-day fix window after launch.

To make a fixed price meaningful, most engagements start with one of two fixed-price, one-week entry offers at $1,500 each: the Scope Sprint for a new product, or the AI Opportunity Audit for AI and automation in an existing business. Each ends with a written scope and a fixed quote, which are yours to keep and to take elsewhere if you wish.

Builds are then priced against that scope: MVP development from $9,800, a single automation from $4,800, a custom AI agent from $9,800. Ongoing work after launch is a monthly arrangement, from $2,500 a month. Details are on the pricing page.

#Bottom line

If you can describe what should be running at the end, ask for a fixed price, and expect to do some scoping work first to earn it. If you cannot, pay for a short, fixed-price discovery until you can. Keep time and materials for research, for unfamiliar older systems, and for continuous work on a live product, and always with a cap. If you would like a second opinion on how to structure a project, book a free 20-minute call.

Common questions.

• Is fixed price or time and materials better for a startup MVP?

  Fixed price, in most cases. A first version has to fit a budget, and the discipline of defining the scope is exactly what an MVP needs. The condition is a proper scoping step first, so the price is based on something real. Time and materials suits a startup better once the product is live and the work becomes continuous improvement.

• Is a fixed-price project more expensive than time and materials?

  The quote is often a little higher, because it includes an allowance for risk that the supplier is now carrying. The final bill is often lower, because overruns are not yours to pay. Time and materials looks cheaper at the estimate stage and costs more when the work takes longer than expected, which software work frequently does.

• What is a not-to-exceed or capped time and materials contract?

  It is time and materials with a ceiling. You pay for the hours actually worked, up to an agreed maximum, and the supplier cannot bill beyond it without your approval. It gives you some of the flexibility of time and materials with a known worst case. Check what happens when the cap is reached with work unfinished.

• What happens when requirements change during a fixed-price project?

  A good agreement has a change process. Small changes are usually swapped in by removing something of similar size. Larger ones are estimated, priced and approved before work starts, or parked for a later phase. Change is normal. The problem is unpriced change, which a clear process prevents for both sides.

• Can you run an agile project on a fixed price?

  Yes. The common way is to fix the budget and the timeline and keep some flexibility in the scope: the must-have outcomes are fixed, and the detail of how each is delivered is refined as you go, with a demo every week or two. Another way is a fixed price per phase or per sprint, agreed one at a time.