topfoc.usStart a build
Ask this question and you get quotes from $1,000 to $300,000, which is another way of saying the question is underspecified. The honest answer is that the price follows three decisions, and you can place yourself on the scale before talking to anyone.
The three decisions that set the price
Who builds it. A technical founder with a starter kit and an AI assistant can get a working SaaS live for under $5,000, most of it tooling and infrastructure. Freelancers typically land between $20,000 and $60,000. Studios and agencies mostly quote $30,000 to $120,000, with US rates around $150 to $200 an hour and Eastern European teams at $50 to $80 for comparable work.
What it has to do. A CRUD app with auth and Stripe checkout is the cheap end. Anything past that moves you up a bracket, and the jumps are bigger than they look on the feature list. On COSMABL, a booking marketplace, the hard parts were not the screens: in-person card payments through Stripe Terminal, geospatial matching of practitioners to clients, and a regulated industry's advertising rules built into the product itself. The single most expensive sentence in a spec is "like X, but", where X is a product that took a funded team five years.
How finished it has to feel. A tool for twelve warehouse operators can look like a spreadsheet and everyone is happy. A consumer product competing for attention cannot. Polish is real work; a designer, motion, empty states, and onboarding can be a third of the budget.
Where the surprises hide
The quote covers the build. The overruns come from the parts nobody put in the spec: migrations when the data model turns out wrong, the admin panel, email deliverability, the second and third payment edge case, and every integration that "has an API" but has an API from 2011. A team that has shipped SaaS before prices these in up front. A team that hasn't, bills you for them in month four.
The other classic is paying twice. A cheap build that has to be rewritten before it can grow costs more than the honest quote you turned down. Building cheap is still fine, as long as you know going in which parts are throwaway.
When each route makes sense
Prompt-built or starter kit, if you can code and mainly need to test whether anyone will pay. The demo does not have to survive success; you learn, then rebuild the parts that matter. Our own learning platform MVP was built in this mode: four weeks from idea to working software, with anything that would not change the build-or-drop decision cut from scope.
Freelancer, if the scope is small and truly fixed, and you can judge code quality yourself or know someone who can. A studio, if the product is the business, the scope will move, and you want one accountable party for design, build, and the launch. That last one is the work we do, so weigh the bias, but the categories hold regardless of who you hire.
What we tell people who ask us
Budget for the smallest thing a real customer would pay for, then add 30% for what you learn on the way. If the number a team quotes you has no assumptions attached, the assumptions are hiding in the invoice. And if two quotes differ by 4x, they are not quoting the same product; make them each list what is excluded, and the difference becomes visible fast.
The case studies show what our own builds have bought people. If you want a number for your specific idea, the first conversation is free and usually settles it.