.webp)
A founder we spoke to last month had spent £42k on an MVP for a corporate expense management tool. Two weeks after launch, first users told her the workflow did not fit how their teams actually handled expenses. A £2k Figma prototype tested with 6 users in week one would have shown this. She built the wrong thing well.
This article is written for anyone at the whiteboard trying to decide between a POC, a prototype, or a full MVP. Clear definitions. Real cost bands. Honest guidance on when each one fits. And the failure modes that turn savings into a well-built wrong thing.
If you are the founder, PM, or innovation director looking at a budget and unsure what to build first, this is written for you.
Clear Definitions: POC, Prototype, MVP
The three terms are commonly muddled together. They mean different things.
POC (Proof of Concept). A technical experiment to prove that something works, or does not. Usually not user-facing. Often no UI at all, just data or output. Answers the question "can we build it?" Examples: a script that shows AI can classify our support tickets accurately; a load test showing our proposed architecture handles 100k concurrent users; a spike verifying we can integrate with a legacy carrier's EDI feed.
Prototype. A user-facing simulation or partial build to test design, workflow, or user response. Two variants:
Low-fi prototype: wireframes, sketches, or Figma click-throughs. Users click through the flow but nothing behind the buttons works.
Hi-fi prototype: interactive Figma or Framer, or a coded front-end with mock data. Feels almost like a real app but has no persistent backend.
Both answer "will they use it?" Nielsen Norman Group's prototyping research shows even 5 to 8 user tests on a prototype surface most of the major usability and workflow issues.
MVP (Minimum Viable Product). A functional shipped product with real users, real data persistence, real payment (if applicable), and real feedback. Answers "will they pay for it?" or "will they retain?" MVP has working backend, working auth, working payments, and is deployed for real users to actually use.
The Y Combinator Library has decades of essays reinforcing this: get the cheapest possible test of your hypothesis before spending on the next expensive test.
What Question Each One Actually Answers
Think of it as three sequential questions.
POC answers: "Can we build it?" Novel technology, unclear technical feasibility, integration risk. If you cannot answer yes, no amount of design or product effort matters.
Prototype answers: "Will they use it?" Assumes the technology works. Tests whether the workflow, design, and user experience match how real users behave. Cheap to iterate. Expensive to skip.
MVP answers: "Will they pay for it? Will they retain?" Assumes technology works and workflow fits. Tests commercial and retention hypothesis with real users, real money, real behaviour.
The mistake most founders make is starting with the MVP question when they have not answered the prototype question yet. They build an app that works, then discover users do not want the workflow. £40k spent, hypothesis untested.
When to Build Each (Decision Framework)
A simple decision framework for what to build next.
Build a POC when:
You are using novel technology (new AI model, unusual protocol, unproven integration)
You have technical risk that would make the MVP moot if it fails
Technical feasibility is unclear or debated on your team
Build a prototype when:
Technology risk is low (standard web app, mobile app, common integrations)
User workflow, design, or navigation is unclear
You are choosing between two or more possible product approaches
You have not tested the concept with real target users
Build an MVP when:
POC and prototype questions are answered
You have signal (user interviews, prototype tests, waiting list) that users want it
You need to test payment, retention, or commercial hypothesis
You have budget to iterate for 6 to 12 months post-launch
Order matters. Some products skip the POC step (standard technology, no feasibility risk). Very few good outcomes come from skipping the prototype step. Almost no good outcomes come from starting with the MVP for a novel product with no user validation.
Real 2026 Cost Bands
POC. £3k to £25k over 2 to 6 weeks. A single engineer or small team validates one technical question. Deliverable is often a report plus code or data showing whether it works.
Low-fi prototype. £500 to £5k over 1 to 2 weeks. Wireframes on paper, whiteboard, or in Figma. Tested with 5 to 10 target users in short interviews.
Hi-fi interactive prototype. £2k to £15k over 2 to 4 weeks. Figma or Framer with realistic content and interactions. Tested with 8 to 15 target users. Often includes a designer plus a light front-end engineer.
MVP. £15k to £80k over 8 to 16 weeks. See our earlier post on how to build a SaaS MVP in 90 days for the detailed sprint plan and team structure.
Total budget across all three stages typically £20k to £120k depending on complexity. The programmes that spend £60k on POC + prototype + tight MVP outperform the programmes that spend £60k on a full MVP alone. Cheap questions answered early cost less than expensive questions answered late.
Common Failure Modes
Three failure modes we see repeatedly.
Building an MVP when a prototype would have answered the question. Founder skips prototype because "we know what to build." £40k later, users tell them the workflow is wrong. The £2k prototype would have caught it in week one.
Skipping a POC on genuinely novel technology. Team is confident it will work. Month 3 of MVP build, the integration fails, the AI model does not generalise, or the load pattern is wrong. Rework doubles the budget.
Confusing hi-fi prototype with MVP. Prototype looks like a real app. Founder shows investors. Investors expect a real product in 8 weeks. Prototype has no backend, no auth, no payment. Twelve weeks of unexpected engineering to actually ship it. Manageable if planned; disastrous if not.
What We See Founders Get Wrong
Across 50-plus shipped products at WhiteStone, three patterns recur.
Wanting to skip validation. Founders in love with their idea often want to skip prototype testing because they "know" users will love it. They are almost always partly wrong. Not because their idea is bad but because real user behaviour surfaces detail no founder can anticipate.
Confusing budget for outcome. Spending more does not answer questions better. It just answers questions with more scope. A £3k prototype tested with 8 real users often produces sharper answers than a £30k prototype tested with 2 friends.
Not writing down the question. Every POC, prototype, and MVP should start with a written hypothesis: "we believe X. If Y happens, we know we are right. If Z happens, we know we are wrong." Without a written hypothesis, teams build stuff and interpret results however they want. See our earlier post on MVP development guide for non-technical founders for the spec-writing pattern.
You can see our shipped work at our portfolio. If you want a candid conversation about which stage fits your specific question, book a product discovery session with WhiteStone.
Frequently Asked Questions
Do I need a prototype before an MVP?
Almost always yes. A prototype tests workflow and design with real target users for 5 to 15 percent of MVP cost. Skipping prototype and going straight to MVP is common; producing a good MVP without prototype validation is rare.
When is a POC the right choice?
When you have genuine technical risk: novel technology, unproven integration, feasibility that the team debates. POC removes that risk cheaply before you commit to prototype or MVP investment. If technology risk is low, skip POC and start with prototype.
How much cheaper is a prototype than an MVP?
Low-fi prototype £500 to £5k, hi-fi prototype £2k to £15k, MVP £15k to £80k. A hi-fi prototype typically costs 5 to 20 percent of an MVP. Cheaper if you use no-code prototyping tools rather than coded front-ends.
Can I show a prototype to investors instead of an MVP?
Yes for pre-seed and seed rounds increasingly. Investors know the difference. A prototype demonstrates product thinking, workflow judgement, and design sensibility. What it does not demonstrate is user retention or paying customers. Match your stage to investor expectation.
What if I do not have budget for all three?
Most products can skip POC (standard technology). Skip prototype at your peril. If budget forces a choice between prototype and MVP, do both smaller versions rather than skip the prototype: £3k low-fi prototype plus £25k tight MVP beats £30k MVP with no prototype validation.
The One Thing to Remember
The three terms are not marketing labels. They answer three different questions and cost three different amounts. The founders who ship well ask cheap questions before expensive ones. The founders who do not ship well spend £60k proving they should have spent £3k first.
If you want a candid conversation about which question your idea should answer first, browse our custom software development services or come to the discovery session.
