Time and materials vs fixed price, and who approves what.

A fixed price contract names one number for a scope both sides have already described, and puts the risk of a bad estimate on the agency. Time and materials names an hourly rate, bills the hours actually worked, and moves that risk to the client, which means nobody knows the total until the work stops. Most comparisons stop there and hand you a table of pros and cons. The part that decides how your year goes is narrower than that, and it's this: under one model somebody has to say yes in writing before extra hours get worked, and under the other they don't.

That difference is invisible for the first month and then arrives all at once, usually the week a client asks for something reasonable that the original number was never sized for. What follows is one nine-week project written both ways, with the same request landing in week six of each, so you can see where the money actually goes and where the record does or doesn't exist.

What's the difference between time and materials and fixed price?

The models differ on when the total gets decided, and everything else follows from that. A fixed price contract settles the number before the work starts, so the scope has to be described well enough to price, and any addition to it needs a formal change to the contract. Time and materials settles the rate before the work starts and the number when it ends, so the scope can stay loose and additions just show up as more hours on the next invoice.

The construction version of this comparison spends a lot of words on material markups, commonly 15 to 35 percent over cost, and on retainage held back until the job is signed off. Almost none of that survives the trip to an agency. Your materials are a few software licenses, some stock photography and a font license, and they usually pass through at cost or close to it. What you're really selling under either model is hours, which is why the choice is less about cost recovery and more about who gets to authorize the next block of them.

The two models also carry very different amounts of paperwork. Fixed price is light to administer and heavy to change. Time and materials is the reverse, because someone has to track and defend every hour, and the client is entitled to ask what each one bought.

Who actually carries the risk under each one?

Under fixed price the agency carries it, and under time and materials the client does. If a $47,200 build takes 40 percent more hours than the estimate assumed, the fixed price version pays the agency $47,200 anyway and the difference comes out of margin. The same overrun on a time and materials contract simply produces a larger invoice, which the client pays, and the agency's margin per hour never moves.

Clients know this, which is why almost every serious time and materials contract arrives with a ceiling attached. A not-to-exceed clause caps the total at the estimate plus a stated percentage, and the agency can't invoice above it without fresh written approval. A guaranteed maximum price does something similar on larger engagements, with the agency absorbing anything over the cap and the two sides usually splitting whatever comes in under it.

Both of those devices give the client budget certainty without taking away the flexibility that made time and materials attractive. They also create a number nobody watches. The cap sits there looking generous in week one and gets consumed by requests that each felt too small to mention, which is the failure mode the rest of this piece is about.

What does the same $47,200 build look like priced both ways?

Take Fernway Outfitters, an outdoor gear retailer that wants a member portal with subscription billing on the front of it, scoped at nine weeks of work. Written as a fixed price contract it's one line: $47,200 for the portal, the billing integration and the two rounds of revisions the scope names, invoiced in three payments against defined milestones.

Written as time and materials it's a rate and an estimate. The blended rate is $160 an hour, the estimate is 295 hours, and 295 hours at $160 is the same $47,200. The client asks for a ceiling, so the contract adds a not-to-exceed figure of the estimate plus 15 percent, which is $54,280. That gives the agency $7,080 of room above the estimate before anybody has to have a conversation.

On paper those two contracts are the same deal, and an agency owner comparing them will usually pick on feel: the fixed price one looks cleaner to send and harder to live inside, the hourly one safer to deliver and harder to sell. Neither instinct tells you what happens next.

What happens when the client asks for something in week six?

Fernway wants gift subscriptions added, so a member can buy a plan for somebody else. It's a fair request, it wasn't in the scope, and the work comes to 34 hours across checkout, the billing integration and the emails. At $160 an hour that's $5,440.

Under the fixed price contract, $5,440 is a change order. The agency prices it, sends it, and nothing gets built until somebody at Fernway with the authority to spend that money says yes in writing. The contract value moves from $47,200 to $52,640 and the new date moves with it. It's a small amount of friction on a Tuesday afternoon, and it's the entire mechanism.

Under the time and materials contract there's no approval step at all, because the hours are billable either way. Somebody picks the work up in the ordinary run of the week, it lands on the next invoice, and the total runs from $47,200 to $52,640, which is still $1,640 under the cap. The work is the same and the money is the same and both versions finish in the same week. What differs is that one of them produced a written yes from a named person, and the other produced 34 hours in an invoice line item next month.

Does time and materials really mean no change orders?

Not for getting paid, and that's precisely why it stings. The hours flow through whether or not anyone approved them, so the disagreement doesn't get avoided, it gets deferred to whenever the client next reads an invoice carefully. On the Fernway job that invoice is 11.5 percent larger than the estimate the finance team budgeted against, and the question that comes back is which of these hours we asked for.

The gift subscription work is easy to defend, because it was one request from one person on one day. What's hard to defend is how the rest of the $7,080 disappeared. On a real nine-week build those 34 hours don't arrive as one request. They arrive as four: nine hours to change how the plan comparison reads, six to add a second currency to the pricing table, fourteen for gift subscriptions, and five for a report somebody wanted for a board meeting. Every one of them was small enough that asking for written approval felt like bureaucracy, and together they used 77 percent of the buffer.

That's the honest case against the flexible model. It doesn't remove the change conversation, it removes the moment when the change conversation was cheap. Where the line falls between the original scope and everything after it is worth settling in writing at the start, which is what the difference between in scope and out of scope is really for.

Which one should an agency pick for a new client?

Fixed price, unless you genuinely can't describe the finished thing. Clients buy outcomes rather than a supply of hours, a single number is far easier for them to get approved internally, and the discipline of having to price the work forces the scope to get specific in ways that protect both sides later. A fixed price you can't defend is usually a scope you didn't write properly, and that's a solvable problem rather than a reason to switch models.

The exception is worth taking seriously. When the requirements depend on decisions the client hasn't made, or on a system nobody has seen inside yet, a fixed price is a guess with a signature underneath it, and you'll spend the project either eating overruns or arguing about them. That's the case for billing by the hour, and it's the same case a good scope of work clients can't misread tries to close before it comes up.

One more consideration, since it decides more of these than anyone admits. Time and materials asks the client to trust your hours before you've earned that trust. On a first engagement with a client who doesn't know you, a fixed number with a written change route is easier to say yes to and easier to be judged on.

Can one project use both?

Yes, and on the projects that are genuinely hard to price it's the better answer. The usual shape is sequential: a paid discovery phase billed at a day rate whose only deliverable is a written specification, then a fixed price for the build that specification defines. The client pays for the uncertainty once, at a small scale, instead of paying for it across the whole engagement.

The other shape is a carve-out. Fix the price on everything you can describe today and name one part as time and materials, such as migrating six years of member data out of a platform nobody has given you access to yet. Give the carve-out its own rate, its own cap and its own approval line, because an unbounded carve-out has a habit of growing until it's the project. The mechanics of wiring that into the paperwork sit with how a statement of work and a contract connect, since the fee structure belongs in one and the change process belongs in the other.

The thirty-four hours nobody had to approve

Both versions of the Fernway job invoiced $52,640 and finished the same week, and only one of them can show who asked for the extra $5,440 and who agreed to pay it. That's the whole argument for fixed price in agency work, and it isn't really an argument about pricing at all. A fixed number is just the thing that forces a written yes at the moment the request is made, when it costs one email, rather than at the moment the invoice arrives, when it costs a relationship.

Which means the model matters less than the habit. An agency that bills by the hour and still sends a priced note before starting anything over four hours has the record. An agency on fixed price that quietly absorbs 34 hours gives away 11.5 percent of the contract and has nothing to point at either. Docket exists for that one moment: the request comes in on a link, you reply with the hours and the dollars, and the client's written yes lands on the project record before the work starts. If the fee section of your scope needs rewriting first, the worked example in the software development scope of work template shows the discovery-then-fixed-price split in full, and a change request form covers the fields each request should carry.

Frequently asked questions