A properly built MVP costs somewhere between $30,000 and $80,000, depending on what it is. Hold onto that number. It is not the $8,000 someone quotes you after a weekend with a no-code tool, and it is not the $130,000 a full custom software project averages either (Clutch). One number is too optimistic. The other is too big for what an MVP is supposed to be.
The single biggest thing that moves the price is who builds it. A US team runs $50 to $99 an hour. A team in India, the Philippines, Ukraine or Mexico runs $25 to $49 (Clutch). Product type is the second lever: a SaaS MVP typically lands at $40,000 to $80,000, a marketplace at $40,000 to $60,000, a mobile app at $30,000 to $50,000, e-commerce at $30,000 to $60,000 (Codica). Two quotes with a three times gap between them usually differ on team location and category, not on honesty.
Inside that total, the split holds up fairly consistently across studios. Discovery and research run $4,000 to $8,000. Design runs $6,000 to $10,000. The core build is $30,000 and up. Quality assurance is $8,000 to $12,000 (Codica). One studio's own worked example, a marketplace build at 1,164 hours and $50 an hour, broke down as $6,800 for design, $2,800 for architecture, $4,800 for integrations, $32,300 for core functionality, and $9,200 for project management and QA. Add 15 to 25 percent on top as a contingency, because the first estimate is never the number you land on.
Minimum was never meant to describe cheap. The term traces back to a 2001 framework from Frank Robinson, later popularised by Eric Ries and Steve Blank: the smallest version of a product that earns honest feedback from real customers before a company commits to building the rest. It is a spending strategy, not a discount. Spend the least amount required to learn whether the idea holds up, then build the rest with an answer in hand instead of a guess.
QA is the easiest line item to cut from a quote, because it happens last and there is no feature to point at when it goes well. It is also the worst place to cut. Skipping it doesn't remove the cost. It moves the cost to launch week, where a bug in front of the first real users costs more in trust than it ever saved in the budget. Skipping discovery to reach code sooner does the same thing. Both feel like savings on the invoice. Both usually show up later as a bigger bill.
A single lump sum with no line for discovery, design and QA is worth a second look before you sign it. It usually means one of two things: the studio has already decided to skip a phase, or those phases are hidden inside a total that will grow once the build starts. A quote broken into those four pieces, even roughly, is the one that tends to survive contact with the real project.
Before you sign anything, ask for the number by phase, not just the total. A studio that can hand you discovery, design, build and QA as separate lines, with a contingency sitting on top, has scoped the work. One that hands you a single figure is asking for trust it hasn't earned yet.