Why five quotes can differ by five times
It is almost never five firms disagreeing about the price of an hour. New
Zealand development rates cluster fairly tightly. The spread comes from five
firms building different things, because the brief let them.
One quoted the screens you described. Another quoted those plus the integration
with your accounting system, migrating fifteen years of existing records,
training your staff, and a year of support.
Both are honest. Only one of them is what you meant.
Make the quotes comparable
State these explicitly in the brief and the spread collapses:
- Integration. Which systems must it talk to, and in which direction?
- Data migration. Is existing data coming across? How much, from where, and
- Admin tooling. Who at your end needs to look things up, fix a record or
- Reporting. What must someone be able to get out of it?
- Training and handover. Who is trained, and what documentation do you
- Support. What happens in the first three months after go-live?
- Hosting. Who provides it and who pays?
Every one of those is a real cost that someone will pay. Leaving them out of the
brief does not remove them; it just means you compare quotes that included them
against quotes that did not.
Try the cheap answer first
Subscription software is inexpensive, available this afternoon, and improved by
somebody else at their expense. Before commissioning anything, spend a fortnight
genuinely trying the closest product with your real data.
You will learn three things: whether the standard model matches how you work,
which two or three things it cannot do, and whether your team will use it. All
three make every quote you receive afterwards more accurate, because you can
describe a gap rather than a wish.
Building becomes the right answer when:
- Your process genuinely does not fit anything available.
- Per-seat pricing has become painful at your headcount.
- The thing you need sits across several systems that will not talk.
- The software is part of what makes you competitive.
Ask every developer which off-the-shelf product they considered and what it
would have missed. A firm that names one and loses the job has just told you
they will be honest later, when it matters more.
Build small, then build more
The most reliable predictor of a good outcome is how quickly something real
reaches the people who do the work.
A project that specifies everything up front is specifying guesses. Two months
of actual use will teach you things no workshop would have surfaced — that the
screen everyone argued about is used twice a month, and the one nobody mentioned
is opened forty times a day.
So ask each firm what they would build first, and how long until someone is
using it. A proposal where nothing is usable for six months is a proposal to
find out late.
What good quoting looks like
The firm you want is the one that asks uncomfortable questions before pricing:
- What happens today when this goes wrong?
- Who does this work now, and what do they do that is not in the process
- What are the exceptions? The urgent order, the customer with different terms,
- What does this cost you a year at the moment?
- What would make you say, in a year, that this was worth it?
That last question is the one that separates a supplier from a partner. If
nobody has asked what success looks like, nobody is aiming at it.
Fixed price, or time and materials?
Both are legitimate and they suit different things.
Fixed price works when the scope is genuinely well defined and unlikely to
change. You get certainty; the developer prices the risk, so you pay for it
whether or not it happens, and every change is a variation.
Time and materials with a cap works when you expect to learn during the
project, which is most of the time. You need to trust the firm and stay
involved, and you should insist on a ceiling agreed in advance and a call before
it is reached.
What to avoid is a fixed price for a scope nobody has examined. It is either
padded against the worst case, or it will be revised in month two.
What it costs to keep
Budget 15 to 20 percent of the build cost each year, plus hosting from around
$50 a month for most business systems. That covers security patches, dependency
updates, the small changes every system needs once real people use it, and
somebody available when something breaks.
Software with no maintenance arrangement becomes a rescue job in about three
years, and the rescue costs more than the maintenance would have.
Getting five quotes
Describe the problem rather than the solution: what happens today, who does it,
how often, what it costs you, and what would have to be true for this to be
worth doing.
Up to five developers can respond, free, with no obligation. Where their
proposals differ is the most useful information you will get — those differences
are the questions you did not know to ask.