Every founder I talk to has a scar from the wrong contract model.
Usually it goes one of two ways. Either they quoted a fixed price, the scope grew, and they ate three weeks of unpaid work because saying "that's a change request" felt petty. Or they billed time and materials, the client watched the hours pile up, lost their nerve halfway through, and now there's a tense call every Friday about why the invoice keeps growing.
Both founders blame the client. Both are wrong. They picked a pricing model that didn't match the shape of the work, and the model did exactly what it was always going to do.
Before starting Scopeyard, I spent six years running a product development studio building software for banks, and the rule held every time: the contract model isn't an admin detail you sort out after the proposal. It is the proposal. It decides who absorbs the uncertainty, and uncertainty is the only thing in a project that's guaranteed.
1. Understand what you're actually choosing between
A fixed-price contract says: I will deliver this defined scope for this number, whatever it takes. The client gets certainty. You absorb every unknown.
Time and materials says: I will bill for the hours I work and the resources I use. You get paid for effort. The client absorbs the unknowns, and the risk that the bill keeps climbing.
That's the whole thing. Everyone dresses it up, but a pricing model is just a decision about who carries risk. And the data on how much risk there is to carry is sobering. Around 70% of software projects exceed their budget, with overruns averaging about 27%. Only 31% of software projects finish on time, on budget, and to expectations. One analysis found that roughly 90% of projects need changes during development.
Read those numbers again, because they decide everything. If nine in ten projects change mid-flight, then a fixed price quoted on day one is a bet you're making against the base rate. Sometimes you win it. Mostly the house wins.
2. Fixed price isn't safer — it just hides the risk in your buffer
Clients love fixed price because it feels safe. Agencies offer it because clients love it. But "safe" for the client means "dangerous for you," and most agencies underprice that danger.
To make a fixed price survive contact with reality, you have to pad it. Industry estimates put the buffer agencies bake into fixed bids at anywhere from 15% to 30% on top of the honest estimate, and sometimes more. So one of two things happens. Either you quote the padded number and the client thinks you're expensive, or you quote the lean number to win the deal and absorb the overrun yourself.
There's a third, uglier outcome the research is blunt about: when a vendor misjudges a fixed bid and costs run over, the temptation is to cut corners or quietly reduce quality to protect the margin. I've seen agencies do it. The client gets a worse product, the agency gets a worse reputation, and everyone pretends the fixed price "worked" because the invoice matched the quote.
Fixed price also turns every conversation adversarial the moment scope shifts. The client asks for a small change. To you it's unbudgeted work. To them it was "obviously included." Now you're negotiating instead of building. The model that was meant to remove friction manufactures it.
3. T&M is honest but it needs a spine
Time and materials fixes the buffer problem. You bill for what you do, scope can flex, and you're never quietly funding someone else's indecision. For anything longer than a few weeks that involves new functionality, design, data integration, or business requirements that are still moving, T&M is almost always the better model on paper.
The catch is trust. A naked T&M contract asks the client to sign a blank cheque, and clients who've been burned won't. The disadvantages are real: the budget isn't fixed, so there's always a chance of overrun, and every hour has to be tracked and justified. Do that tracking badly — a vague invoice, no burn-down, surprise totals — and you'll trigger exactly the Friday-afternoon panic call I mentioned at the top.
T&M doesn't fail because it's a bad model. It fails when agencies run it without transparency. The hours are real; the problem is the client can't see them accumulating, so the total feels like a shock instead of a story they watched unfold.
4. The hybrid most good agencies actually run
The framing of "fixed vs T&M" is a trap, because the best answer is usually neither in its pure form. The model I'd default to for most agency work is capped time and materials — sometimes called a "not to exceed" agreement.
You bill T&M, so you're paid for real effort and scope can flex. But you agree a ceiling the invoice won't cross without a fresh conversation. The client gets the budget protection that makes fixed price feel safe. You keep the flexibility and honesty that make T&M sustainable. The risk gets shared instead of dumped entirely on one side.
Here's how I think about matching model to project:
| Project shape | Best model | Why |
|---|---|---|
| Small, crisp, fully specified (a landing page, a defined integration) | Fixed price | Low uncertainty, so the buffer is cheap and certainty wins the deal |
| Medium build, scope mostly known but evolving | Capped T&M | Flex for the unknowns, a ceiling for the client's nerves |
| Large or exploratory, requirements still forming | Pure T&M with tight reporting | Too much uncertainty to cap honestly; transparency is the protection |
| Ongoing work, no fixed endpoint | Retainer | Recurring scope doesn't fit a one-off price at all |
Notice the logic running down the table: the more uncertainty in the work, the further you move from fixed price. You're not picking a model you prefer. You're reading where the unknowns sit and choosing accordingly.
5. Price discovery so you can actually offer fixed price
Sometimes the client genuinely needs a fixed number: a board approval, a grant, a hard budget. Fine. But you can only quote a fixed price you'll survive if the scope is genuinely known, and on day one it never is.
This is the Cone of Uncertainty, Barry Boehm's idea from 1981: at the start of a project, before requirements are clear, estimates can be off by as much as 4x in either direction — a roughly 16x spread between your best and worst case. As you learn, the cone narrows to something like ±10–15%.
So the move is simple. Don't quote a fixed price from the wide end of the cone. Run a paid discovery phase first — map the requirements, integrations, edge cases and the non-functional work everyone forgets — and then quote fixed against a scope you can actually see. Charge for discovery as its own small engagement. A client who won't pay a few thousand to de-risk a six-figure build is telling you something useful about how the rest of the project will go.
6. Whatever model you pick, make the work visible
The real differentiator between an agency that grows on T&M and one that gets fired on it isn't the rate. It's whether the client can see what they're paying for as it happens.
A fixed-price project hides the work behind a single number, which is fine until scope shifts and you're suddenly arguing about what "done" meant. A T&M project that surfaces every approved deliverable, every hour, every decision the client signed off on turns the invoice into a record of progress instead of a demand for trust.
This is the bit most pricing arguments skip. The model sets who carries the risk; the delivery system decides whether anyone panics about it. We built Scopeyard partly around this — every deliverable carries a tracked client approval, so when you bill, the sign-off is already on record and "I never agreed to that" stops being a conversation you have. If you run product work, the way I think about this is laid out more fully on our page for product agencies. The model you quote and the system you deliver in are the same decision viewed from two ends.
7. A worked example
Say a build is honestly 400 hours at a $120 blended rate — $48,000 of real work.
Quote it fixed and you'll pad it. At a 25% buffer that's $60,000. Win the deal and the client paid $12,000 for your uncertainty. Misjudge the scope and run to 480 hours, and you've handed back the buffer and then some, with a stressed team and a tempting shortcut on quality.
Quote it capped T&M at $120/hour with a $55,000 ceiling and transparent weekly reporting. If it lands at 400 hours, the client pays $48,000 — less than the fixed quote — and trusts you more for it. If it runs to the cap, you have a documented trail of approved work explaining exactly why, and a clean conversation about the bit beyond it. Same project, same rate. The model decided who slept badly.
Final thoughts
Stop asking which pricing model is better in the abstract. There isn't one. Fixed price is a tool for low uncertainty, T&M is a tool for high uncertainty, and capped T&M is the honest default for the messy middle where most agency work actually lives.
The agencies that get burned aren't the ones who picked "wrong." They're the ones who let the client pick for them, then acted surprised when the model did its job.
Price the uncertainty, not the optimism.