What does an MVP actually cost to build?

A working MVP costs $10,000 to $30,000 NZD for most products, built with a modern AI-assisted process. A genuinely complex multi-sided product runs around $50,000. If you've been quoted $100,000 or more for a first version, you're usually not looking at a market rate. You're looking at a scoping failure.

The build is now the cheap part

Here's what's changed, and a lot of pricing hasn't caught up with it. AI has collapsed the cost of actually writing software. What took a four-person agency team three months in 2022 takes one experienced builder a few weeks now, and the quality gap has closed faster than most people buying software realise.

Which means the question "what does an MVP cost" has quietly become a different question: how good is your scope? The expensive skill isn't producing the code anymore. It's deciding what the first version must do, what it must not do yet, and which parts you should buy off the shelf instead of building at all. Get those decisions right and the build itself is the cheap part. Get them wrong and no hourly rate will save you.

What actually drives the price

Three things move an MVP quote more than everything else combined:

User types. A tool with one kind of user is one product. A marketplace or platform with three kinds of users is three products that have to agree with each other. Every side you add roughly multiplies the surface area.

Integrations. Each external system, like payments, accounting, calendars, or suppliers, is its own small project with its own edge cases. Four integrations can cost more than the core product.

Platforms. Web only is one build. Web plus mobile is closer to two.

Rough bands, NZD, AI-era process: a single-sided internal tool or simple SaaS lands around $10,000 to $15,000. A standard product with payments and a couple of integrations, $15,000 to $30,000. A multi-sided platform with real integration depth, around $50,000.

What $50k buys: My Wellness

The top of that curve isn't theoretical. My Wellness was a build for a Queenstown naturopathic clinic, and it was genuinely complex: a three-sided product with a founder interface, a practitioner interface, and a client interface, across web and mobile. Under the hood it integrated Stripe for payments, Xero for accounting, Cal.com for bookings, and native connections to NZ suppliers, with Retool as the front end and Xano as the back end.

That build ballparked around $50,000 NZD. A few years ago, the honest agency quote for the same scope would have started with a 2.

Two things made the number work. The AI-assisted process, obviously. But just as much, the buy-versus-build discipline: Retool and Xano meant the admin panels, database, and plumbing were bought, not built, so the budget went into the parts that were actually specific to the clinic. Spending custom-build money on commodity features is the most common way MVP budgets die.

/ From the workshop · Health platform · Queenstown

My Wellness: 10+ admin hours saved per practitioner every week.

Read the case study →

Why a $100k first version is a scoping failure

A first version exists to answer a question: does this work in the market? That's it. That's the job.

If answering the question costs six figures, you're not asking anymore, you're betting. And almost every six-figure MVP quote I've reviewed contains the same padding: a second user type that could wait, a native app that could be a web app for a year, a custom admin panel that Retool replaces for $100 a month, and integrations for partners who haven't signed anything yet.

The fix is rarely a cheaper builder. It's a smaller question. One user type, one platform, the boring parts bought, and a clear definition of done. You can always spend the second $50,000 once the first one has proven something.

One caveat so the number stays honest: the MVP price is the entry cost, not the whole journey. Software that ships and then sits still starts decaying the day it launches. Budget something for the system to keep learning after launch, because the products that win their category are never the ones that stopped at version one.

The takeaway

The market rate for an MVP is $10,000 to $30,000 NZD, and the spread between that and the six-figure quote is almost never the quality of the code. It's the quality of the scoping. Decide what question your first version is answering, make it as small as the question allows, and the build becomes the cheapest part of your startup.