July 1, 2026

How to vet MVP development companies without getting burned: 12 questions that filter the field

Twelve questions I ask before I let anyone touch my MVP. Skip these and you'll pay $80k for a demo that dies the week after launch.

KEY TAKEAWAYS

↳Ask who writes the code and who owns it on day one. If those answers are vague, walk.
↳A real MVP shop scopes in weeks, not quarters, and shows you the last three things they shipped with URLs.
↳Fixed-bid contracts hide the risk, they don't remove it. Time-and-materials with a hard cap works better.
↳Handoff is the part everyone lies about. Ask exactly what you get on the last day.

I got a call last month from a founder who paid a shop in Bangalore $62,000 for an MVP. The app worked. Kind of. It was a React Native build with no tests, a Firebase backend wired directly to the client with permissive rules, and a Stripe integration that stored card tokens in plaintext in Firestore. The team had disappeared into the next contract. He wanted to know if we could fix it.

We could. It cost him another $28k and six weeks. He would have saved all of it by asking better questions up front.

So here is the list. Twelve questions I would ask any MVP development agency before I signed anything. These are not gotchas. A good shop will answer them fast and want you to ask. A bad one will get squirrely around question four.

1. Who actually writes the code, and where do they sit?

A lot of MVP development companies in the USA and UK sell you senior engineers on the pitch call and staff you with juniors from a subcontractor once the ink is dry. Ask for the names and GitHub profiles of the humans who will touch your repo. Ask what time zone they work in. Ask if they are employees or contractors of the agency. If the answer is “we have a global talent pool,” that means nothing. You want names.

2. Show me the last three things you shipped, with live URLs

Not case studies. Not screenshots. URLs I can load right now. If the shop is a startup MVP development company that only builds internal tools, fine, but I want to see one and click around. Every agency has three projects they are proud of. If they cannot produce them in an email within a day, they either did not build what they claim or they are hiding NDAs so aggressive that you will not be allowed to show your own product either.

3. What is your definition of an MVP?

I have heard “MVP” used to mean a Figma prototype, a landing page with a waitlist, and a full production SaaS with billing and admin dashboards. Get the definition in writing before anyone quotes a price. My working definition: an MVP is the smallest thing that lets a real user do the core job and pay you if the job is worth paying for. Anything else is a demo.

4. What is your stack, and why?

If they say “we are stack agnostic,” they are lying or unfocused. Every shop has a default. Ours is TypeScript, Next.js or Astro on the frontend, Postgres on Hetzner or Supabase, Cloudflare Workers where they fit, Stripe for payments, Resend for email, Turnstile for anti-abuse. That is boring on purpose. Boring ships. If a shop wants to build your MVP on a bespoke Rust backend with a custom ORM, ask who maintains it after they leave.

5. Fixed bid or time and materials?

Fixed bid feels safe. It is not. Fixed bid means the shop has priced in every possible scope change, so you overpay on the happy path, and when you inevitably want to change something, you get nickel-and-dimed on change orders. I prefer time and materials with a hard cap and weekly burn reports. You see the money move. You can stop any Friday.

6. What does the first two weeks look like?

A competent MVP development agency should be able to describe week one and week two in concrete deliverables. Auth working by end of week one. Core data model and one end-to-end flow by end of week two. If the answer is “discovery and alignment,” that is a month of Miro boards and no code. Run.

7. Who owns the code, the domain, the accounts?

Get this in the contract, not the pitch deck. You own the code from commit one. You own the GitHub org, the Vercel account, the Stripe account, the Cloudflare zone, the Apple developer account, the Play Store listing. The agency is a collaborator on your accounts, not the other way around. I have seen founders locked out of their own App Store listing because the agency owned the Apple ID.

8. What happens on the last day?

Ask exactly what you get at handoff. A good answer includes: the repo with a README that a new engineer can follow, a running production environment with documented credentials in a password manager you own, a short Loom walking through the architecture, and a two-week warranty for bugs that surface in production. If the answer is “we send you the code,” that is not handoff, that is abandonment.

9. Can I talk to a founder you shipped for eighteen months ago?

Recent references are easy. Anyone can produce a happy client from last month, before the honeymoon ends. I want to talk to someone whose product has been in the wild for a year and a half. Did the code hold up? Did they need to rewrite it? Did the shop pick up the phone when a Stripe webhook broke at 2am?

10. How do you handle AI features?

If your MVP has any AI in it, and in 2025 most do, ask about model choice, cost controls, and evals. A bespoke MVP development company that hardcodes gpt-4o everywhere with no fallback and no token budget is going to hand you a $9,000 OpenAI bill the week you get on Product Hunt. Ask about Anthropic vs OpenAI vs open models on FAL or Replicate. Ask what happens if the model returns garbage. If you want more on the agent side of this specifically, I wrote about how we build custom agents.

11. What is your bug rate, and how do you know?

Any shop that says “we don’t really have bugs” is either lying or not shipping. I want to hear about their last three production incidents and what they did about them. I want to know if they run any tests at all, even smoke tests. Not 100% coverage, that is a myth for MVPs. Just enough that the auth flow and the payment flow do not silently break.

12. Will you tell me not to build something?

This is the one that separates the mercenaries from the partners. A startup MVP development company that will happily build whatever you ask for, exactly as specified, will make you poorer. You want a team that will look at your Notion doc and say “this feature costs three weeks and no user has asked for it, cut it.” If they never push back, they are treating you as a wallet.

A short note on geography

People ask me if they should hire in the USA, UK, Canada, India, Poland, Argentina. Honest answer: it matters less than the team. I have worked with brilliant shops in Bangalore and terrible ones in Toronto. The questions above filter on quality, not zip code. What geography does affect is time zone overlap and legal recourse. If you are in New York and your team is in Chennai, you need one person on their side who is awake when you are, and you need to accept that suing them if things go sideways is not really an option.

If you want a second set of eyes on a proposal you are about to sign, or you want to talk about scoping an MVP without the theatre, get in touch.

FREQUENTLY ASKED

Common questions

▸What does an MVP actually cost from a reputable agency in 2025?

For a real MVP with auth, one core workflow, payments, and a small admin, expect $25k to $80k in North America or the UK, $10k to $40k in Eastern Europe or Latin America, $6k to $25k in India. Anything under those floors is either a template rebrand or a team that will disappear. Anything way over is discovery theatre or a mismatched scope. Get a weekly burn rate, not a lump sum.

▸Is it better to hire an MVP development company in the USA, UK, or India?

Team quality varies more within a country than between countries. What actually changes with geography is time zone overlap, communication style, and legal recourse. Pick the region where you can get two hours of live overlap per day with the people writing code, and where you trust the contract to mean something. Everything else, ask the twelve questions.

▸How long should an MVP take to build?

Six to twelve weeks of build time is the healthy range for a focused MVP. Under six weeks and you are probably shipping a prototype. Over twelve and scope has crept, or the team is not senior enough to make decisions quickly. If a shop quotes you six months for an MVP, they are quoting you a v1 product, not an MVP, and the price will reflect that.

▸What are the red flags in an MVP agency pitch?

Vague team composition, no live URLs of past work, fixed-bid quotes with no scope document attached, ownership of accounts under the agency name, no defined handoff, promises of a specific launch date on the first call, and an unwillingness to say "cut that feature." Also, decks with more logos than case studies. Logos are cheap to add to a slide.

▸Should I hire an agency or build with an in-house team?

An agency wins when you need to ship in under three months and do not yet know if the product deserves a full-time team. In-house wins when you have product-market fit and the code is now the business. Many founders hire an agency for v0, then hire a lead engineer to take over the repo. That path works if the agency writes handoff-friendly code, which is why questions seven and eight above matter so much.

6/1/2026