Fixed price moves the risk of a bad estimate onto the builder, and you pay a margin for that. Hourly moves it onto you, and you pay only for the hours used. Fixed price means something only when there is a written scope, a change process with a price on it, and money held back until acceptance. I quote fixed, half on signing and half on acceptance.
Who carries the risk in each pricing model?
Both models bill for the same work by the same person. The difference is who pays when the estimate turns out to be wrong, and the estimate is always a bit wrong.
Under a fixed price, the builder carries it. Overrun comes out of their margin. That has two effects worth knowing about: the quote includes a buffer for the overrun that might happen, and the builder now has a financial reason to argue about scope. Neither of those is dishonest. Both are predictable, and both are why the written scope matters so much.
Under hourly, you carry it. Every extra hour is yours. The upside is real: if the job goes well you pay less, and the builder has no reason to fight you over a small addition because they get paid for it either way. The downside is that your exposure has no ceiling, and the only person who can see the meter running is the one holding it.
What must a fixed scope contain?
Plenty of quotes say "fixed price" and mean "a number I said on the phone". A fixed price is only worth having when three things sit behind it.
- A written scope. What gets built, what the software does on the day it lands, and what has been left out on purpose. The exclusions are the half people skip, and the half that gets fought over in week five.
- A change process with a price attached. Requirements move on every build. A change should be a decision you make with a number in front of you, not an argument about whether it was implied in the original conversation.
- Money held back until acceptance. If the whole fee is paid before you have seen the software work, the fixed price has stopped protecting you. Something has to still be owed at the point where you decide whether it does the job.
Ask for all three in writing before the first invoice. If a builder will not put the exclusions on paper, the price is not fixed in any sense that helps you, and that is one of the twelve questions worth asking before you hire anyone.
When is hourly pricing the fair answer?
There are jobs where asking for a fixed price is asking the builder to guess, and the guess gets priced defensively. Three of them:
- Discovery. Working out what to build is the work. Nobody can scope it in advance, because scoping it is the deliverable.
- Systems no one can describe. A database built by someone who left in 2019, an Access file, a Xero setup with six years of habits in it. The unknown is the whole job.
- Small, open-ended maintenance. Ad hoc fixes on something already running, where writing the scope costs more than doing the work.
The sensible shape is both, in order: a short paid discovery step, then a fixed price for the build that comes out of it. That is what a Workflow Review is, at $3,500, with the written plan yours to keep whoever builds from it.
Before assuming custom is the only path, check what a bought tool costs at your size. Abtrac's pricing is a good example of a published per-user schedule worth holding a custom quote against. Legal practice management runs the opposite way: LEAP's New Zealand site publishes no schedule at all, which is exactly the case where a fixed-price custom quote needs to be measured against a demo booking, not a rate card.
How does the same build compare under both pricing models?
Here are the assumptions, stated so you can argue with them. One automation: quotes flowing from the job tool into Xero without anyone retyping them. Estimated at 40 hours of work. Quoted fixed at $6,000, the middle of the $4,000 to $8,000 band on the services page. To compare like with like, the hourly rate is set at the rate that quote implies, $6,000 divided by 40 hours, which is $150 an hour.
| How it goes | Hourly at $150 | Fixed at $6,000 | Who is better off |
|---|---|---|---|
| Lands on estimate, 40 hours | $6,000 | $6,000 | Neither, identical |
| Runs over, 55 hours | $8,250 | $6,000 | You, by $2,250 |
| Comes in early, 30 hours | $4,500 | $6,000 | The builder, by $1,500 |
Read the third row honestly, because it is the cost of the certainty in row two. A fixed price is a bet on the estimate, and the builder prices the bet to win it more often than they lose. What you are buying is a number you can plan around, and a supplier whose incentive is to finish rather than to keep going.
The row that decides it is the second one. On the builds I have run, overruns are more common than underruns, because the surprises live in other people's systems rather than in the code.
Now add the acceptance mechanic. Half on signing and half on acceptance splits that $6,000 into $3,000 and $3,000. If the software never does what the scope says, your exposure is $3,000 rather than the full $6,000, and the builder still has $3,000 of reason to finish it. An hourly arrangement has no equivalent, because every hour is already paid for.
Why does Kove price this way?
Kove quotes fixed scope for builds, four to eight weeks to working software, half on signing and half on acceptance, with a 60-day warranty after launch: anything that breaks in the first sixty days I fix at no charge. There is no retainer wrapped around that warranty.
The reason is simple enough. An owner of a five to twenty person business cannot carry an open-ended software bill, and should not have to sit there wondering whether asking a question is going to cost them $150. A fixed number lets them decide once. It also puts the estimating risk on me, which is where it belongs, because I am the one who knows how long these take.
The exception is the step before the build. A Workflow Review is a fixed fee for a fixed piece of thinking, and it exists so that the build quote is based on a mapped process rather than a hopeful one. Bands for what the build itself lands at are in what custom automation actually costs a Canterbury trades business: $4,000 to $8,000 for a single automation, $10,000 to $22,000 for a custom CRM or job-tracking system.
When is the pricing model not the problem?
Two situations where this whole argument is a distraction.
The first is a job too small to be worth scoping. Under a few hundred dollars, hourly is fine and paperwork costs more than the work. No one needs a change process to add a column to a report.
The second is a build that should not happen at all. If the process changes every few months, no pricing model saves you, because you pay to change the software each time it moves. And if the hours are not there, the model is irrelevant. Run the numbers before the negotiation: the ROI calculator takes the hours a week, what the hour costs you and a build price, and gives you the payback in months. If that comes back past three years, the pricing model was never the question.
Get a number you can plan around
Book a free 30-minute call. Describe the process, and I will tell you whether it is scopeable today or needs a discovery step first. If it needs discovery, I will say so rather than quote a fixed price on a guess.
Questions about pricing a build
Is fixed price or hourly better for custom software in New Zealand?
Fixed price is better when the work can be described before it starts, because it moves the risk of a bad estimate onto the builder and you know the number before you commit. Hourly is fairer when nobody can describe the work yet, which usually means discovery or a system no one in the building understands. The common answer is both, in sequence: a short paid piece of hourly or fixed-fee discovery, then a fixed price for the build that comes out of it.
What does fixed scope actually mean?
Three things, and a price with none of them is just a number attached to a conversation. A written scope listing what gets built and what has been left out on purpose. A change process with a price on it, so a new requirement is a decision rather than an argument. And money held back until acceptance, so the builder still carries risk after the first invoice. Kove uses half on signing and half on acceptance, and the second half only moves once you accept the software.
Why is a fixed price higher than the hours would suggest?
Because it includes the risk of the estimate being wrong. On a build estimated at 40 hours and quoted at $6,000, the implied rate is $150 an hour. Run 15 hours over and the builder absorbs $2,250 of that. Finish 10 hours early and the builder keeps $1,500. You are buying certainty and paying a margin for it, which is a fair trade when the scope is written down.
When should I refuse a fixed price?
When nobody can describe the work yet. A fixed price on a vague scope is worse than hourly, because the builder prices for the worst case and then defends the scope line by line for six weeks. If the systems involved are undocumented or the process changes every few months, pay for a short discovery piece first and fix the price on the far side of it.