MVP vs Prototype vs POC — what you're actually buying

The three things people mean when they say 'MVP'

Ninety percent of the time when a client says 'MVP', they mean one of the other two — usually a prototype. That's fine; the labels matter less than knowing what question you're answering and what the deliverable actually is. Here's the honest breakdown.

• Proof of Concept (POC) — Answers a technical feasibility question. Example: 'Can our proprietary matching algorithm handle 10,000 records in under 200ms?' Deliverable: a stripped-down technical spike, often without a UI. Cost: $5K–$25K. Timeline: 1–3 weeks. Audience: your CTO and one investor.
• Prototype — Answers a design and desirability question. Example: 'Do users understand this workflow and want to complete it?' Deliverable: a clickable Figma or a lightweight interactive build with mocked data. Cost: $15K–$50K. Timeline: 2–5 weeks. Audience: potential customers in user interviews.
• MVP — Answers a business-hypothesis question with real users doing real work. Example: 'Will 50 dentists pay $99/month to automate their insurance claims?' Deliverable: a real product with real auth, real data, real payments. Cost: $40K–$120K. Timeline: 8–14 weeks. Audience: 50–500 paying (or actively-using) users.

How to know which one you need

Ask yourself what question you'd need to answer to justify the next round of funding, or the next $500K of internal budget. If the answer is 'we don't know if we can technically build the hard part' — you need a POC. If it's 'we don't know if the design makes sense to a user who's never seen it' — you need a prototype. If it's 'we don't know if anyone will pay for it' — you need an MVP. If it's more than one of these, do them in order, not simultaneously.

The mistakes we see most often

• Building an MVP when you needed a prototype — $100K burned, six months lost. The product technically works but users can't figure out how to use it. A $30K prototype would have caught the design failure in three weeks.
• Building a prototype when you needed a POC — The design tests great — until engineering discovers in month four that the core algorithm can't be built at cost. Should have spiked the feasibility first.
• Calling everything an MVP so it sounds serious — Fine for pitching, dangerous for scoping. If you tell your vendor 'we need an MVP' and they scope 14 weeks of production engineering when you actually needed a 3-week prototype, that's on both of you.
• Skipping the POC on a hard technical claim — If your pitch deck says 'we do X in Y time' and X is genuinely novel, spike it in code first. Nothing kills a fundraise faster than the technical demo not matching the slide.

What each one should actually cost in 2026

The prices below assume a competent U.S.-based team. Nearshore/offshore is cheaper but the coordination overhead often eats the savings on projects this small. For POCs and prototypes specifically, on-site or same-timezone senior engineers move much faster because the feedback loop is 30 minutes, not 30 hours.

• POC: $5K–$25K — One senior engineer, one to three weeks, one clearly-scoped feasibility question. Anything more expensive isn't a POC — it's a prototype in disguise.
• Prototype: $15K–$50K — One designer plus a light-code engineer, or a senior full-stack solo. Two to five weeks. Emphasis on UX, not code quality — this is going to be thrown away, and it should be.
• MVP: $40K–$120K — Three to five people (design, front-end, back-end, part-time PM), eight to fourteen weeks. Real auth, real data, real deploy. The code should be production-quality because you'll be building on it if the hypothesis validates.