There is no contract model that protects you from a bad partner, and no model that a good partner cannot work within. What the model does decide is who carries the risk of being wrong about scope — and scope is always partly wrong.

Fixed price: you buy certainty and pay a premium for it

The vendor absorbs estimation risk, and prices that risk in. Expect a 20–40% buffer on any honest fixed-price quote. That is not padding — it is the cost of guaranteeing a number against unknowns.

It works well when the scope is genuinely knowable: a marketing site, a defined integration, a migration with a documented source system. It works badly for products, because the moment you learn something from a user, changing course requires a change order, and change orders create an adversarial dynamic exactly when you need to move fast.

The failure mode to watch for: a fixed price that is suspiciously low. The vendor has either misunderstood the scope, or intends to recover margin through change orders. Both end the same way.

Time and materials: you keep flexibility and carry the risk

You pay for effort. If the estimate was optimistic, that is your problem. In exchange you can change direction any week without renegotiating a contract.

T&M is honest when it comes with a cap, a weekly burn report and a scope you both track. It becomes a blank cheque when it comes with none of those. The distinction is not the model — it is whether you can see, every week, what you bought.

Ask for: a not-to-exceed ceiling per phase, a weekly summary of hours against outcomes, and the right to stop at the end of any two-week cycle without penalty. A partner confident in their throughput will agree to all three.

Dedicated team: you buy capacity, not deliverables

A fixed monthly cost for a fixed set of people. You direct the work. This is the model most closely resembling hiring, without the hiring.

It is the right model once the product is live and the work is continuous — support, iteration, scaling, incidents. It is the wrong model before you know what you are building, because you will pay for capacity you cannot yet direct.

The thing to verify: are these named people who stay, or a rotating pool? Continuity is the entire value of the model. If engineers rotate quarterly, you are paying dedicated-team prices for the knowledge loss of an agency.

What we recommend, and why

For most product work the sequence beats any single model:

1. A fixed-price discovery — two to three weeks, priced independently, producing a written scope, architecture and a defensible estimate. You own the output whether or not you continue with the same team.
2. A fixed-price first phase against that scope. Now the number is credible, because it was built on real analysis instead of a sales call.
3. A dedicated team once the product is live and the work becomes continuous.

This sequence puts risk where it can actually be managed at each stage, and it gives you two natural exit points where leaving costs you nothing.

Five clauses that matter more than the model

• IP assignment on payment, not on project completion — so a dispute never holds your product hostage.
• Infrastructure in your accounts, under your billing, from the first commit.
• Written acceptance criteria per deliverable, so "done" is not a matter of opinion.
• A defined exit: handover documentation, credential transfer and a support window, all specified before you need them.
• Named key personnel, with notice required before they are replaced.

The model decides who pays for surprises. The clauses decide whether you can walk away when one arrives.

Every LaunchWe engagement starts with a fixed-price discovery and assigns IP on day one, whichever model follows. If you are comparing proposals right now, we are happy to review the contract terms with you — no obligation to work with us.