Fixed price software development: the number before the call

An agency that will not name a starting figure is asking you to sit through a discovery call to find out whether you are in the right room. Three agencies, three calls, three hours, and you still have nothing you can put in a memo.

So here are the figures first, then the mechanism that turns them into a fee, then the cases where this model is the wrong shape for your problem.

The starting figures, before the argument about them#

Our eleven services each publish a starting figure and a week range. Three of them, so you can calibrate immediately:

Service	From	Timeline
MVP development	$18,000	6 to 10 weeks
Full stack applications	$30,000	10 to 16 weeks
CRM data migration	$4,500	2 to 6 weeks

Those numbers do two things a discovery call cannot. They tell you inside ten seconds whether your budget and the work are in the same universe, and they give you something to write down when your CFO asks what this costs.

Most pages that rank for this term argue for fixed price as a philosophy and never publish a single number. That is the gap this post exists to close.

What "fixed" actually binds#

A starting figure is not a fee. It is the floor of a range, and the honest version of this model has to say what closes that range.

The fee gets fixed at the end of a paid scoping week: one week, $1,500, credited in full against the project. What comes out of it is a fixed fee, a fixed date, a deliverable list and a named owner.

After that, one rule governs the number:

The fee does not move unless you change the scope in writing.

That sentence is doing more work than it looks. It means a mid-project discovery on our side is our problem, not a change order. If the integration turns out to have an undocumented rate limit, or the data is dirtier than the profile suggested, the estimate was wrong and the estimate was ours.

It also cuts the other way, which is the part most agencies leave out. If the scope turns out smaller than we assumed, the fee goes down. A fixed price that can only ever go up is not a fixed price, it is a floor with marketing on top.

Fixed price versus time and materials#

The two models fail differently, and which failure you can tolerate is the actual decision.

Time and materials puts the risk of a bad estimate on you. It is the right model when the scope genuinely cannot be known in advance, for example an open ended research project, or a team you are augmenting for a year. It is the wrong model when the vendor knows more about the work than you do, because then you are paying for their learning curve at your hourly rate.

Fixed price puts the risk of a bad estimate on us. That is only fair if we are allowed to spend real time estimating, which is exactly what the scoping week buys, and if we are allowed to say no. A fixed price quoted off a 30 minute call is not a commitment, it is a lottery ticket with your project as the prize.

The tell for a fixed price you should not trust: it was quoted before anyone looked at your data, your codebase or your portal.

Where the number comes from#

Three things move a figure inside its range, and all three are knowable in a week.

How much of the system already exists. Building a checkout on top of a working catalogue is a different job from building both. The scoping week reads the code rather than the description of the code.

How clean the data is. A migration quote depends on duplicate rate, association integrity and whether the source system has a stable external ID. We profile it before quoting, which is why a data migration can start at $4,500 and still be honest.

How many systems have to agree. One system is a build. Two systems is an integration, with a contract, a backfill and a reconciliation step. Three is a program. The jump between one and two is the largest single cost step in most projects.

What this model is bad at#

It is worth being direct about the cases where you should not want a fixed price from us.

If you want to change direction every two weeks based on user feedback, fixed scope is friction. A scoped fee assumes the destination holds still long enough to reach it.

If the work is genuinely continuous, a fixed project fee is the wrong instrument and so are we. At some point the right answer is a person on your payroll, and the honest comparison between those two options is its own post with the salary assumptions shown.

And if the answer is that you should buy software rather than build it, we will say so. That conversation is cheaper for everyone than a well executed $30,000 build of something that already exists for $200 a month.

What to ask any agency that quotes you#

Four questions, and the answers tell you more than the number does.

1. What is your starting figure for work of this shape, before we talk?
2. What has to happen before that figure becomes a fee?
3. Under what conditions does the fee change, and who decides?
4. If the scope turns out smaller, what happens to the price?

An agency that cannot answer the fourth one has told you the direction their number travels.

Frequently asked questions#

Is fixed price more expensive than hourly?#

Usually yes on paper, because the fee carries the risk of a bad estimate. The comparison that matters is not fee against fee, it is fee against the hourly project that ran 40% long, which is the outcome hourly billing has no mechanism to prevent.

What happens if the project takes longer than you estimated?#

We absorb it. That is what the fee is for. The only thing that moves the number is a scope change you asked for and agreed to in writing.

Can I get a fixed price without the scoping week?#

Not one worth relying on. Anything quoted before we have read your code or profiled your data is a guess, and a guess large enough to be safe is a guess you should not have to pay for.

Do you take equity or revenue share instead of a fee?#

No. Fixed fee only. It keeps the incentive simple and it means you own the result outright, code in your repository and infrastructure in your accounts.