Build a real MVP in 4 weeks

Week 1: ruthless prioritization

You have a list of 20 features you want to build. Only 3 matter in the MVP. Your job this week is to figure out which three.

Start with your core assumption: "Founders will pay for a tool that does..." If your assumption is wrong, everything else fails. The MVP tests that assumption. Nothing more.

For each feature, ask: (1) Does it test the core assumption? (2) Can users accomplish the core action without it? (3) Can it be added after launch?

If the answer to (2) or (3) is yes, cut it. Every feature you cut saves a week and $5,000–10,000.

Week 2: design and user flows

Don't build anything yet. Design the user journey for each core feature. Use wireframes, not finished mockups: wireframes are fast and easy to iterate.

Focus on the happy path. The user signs in, does one thing, sees a result. Onboarding should be 3 screens, not 10. Test your design with 5–10 potential users. Show them the wireframes and let them try to use them without your help.

Iteration is normal. You'll redesign 2–3 times. That's fine: design iteration is cheap. Code iteration is expensive.

Week 3: build the core

You have one week to build. Not perfect, not beautiful: working. Start with a single database table. Users sign in, create one thing, see it in a list. That's it.

Use a template if one exists. Don't build auth from scratch: use Supabase Auth. Don't build a file uploader: use Supabase Storage. Every tool you integrate saves 2–4 days of building.

Skip the polish. Gray buttons are fine. Content can be raw. Animations are unnecessary. Your goal is "works," not "launch-ready."

Deploy early (by day 3). You'll only find bugs once you deploy. Vercel is free and takes minutes.

Week 4: test and iterate

You have 4–5 days. Invite 20–30 people to use your MVP. Watch them. Where do they stop? What do they ask about? What frustrates them?

Take notes. Don't defend your design. If 5 people stop at the same step, that's a signal to change.

By the end of week 4, you have a working product and real user feedback. Launch it. Put it in front of 100+ people. Do they sign up? Do they use it? Do they pay?

If no one cares, you learned that in 4 weeks instead of 6 months. That's a win.

The tech stack for speed

Your choice of technology determines your speed. What works: Next.js (frontend and serverless backend), Supabase (auth + database + storage), Stripe (payments), Vercel (hosting).

Why? Each tool is built for speed. You can authenticate a user in 10 lines of code. A full sign-up flow takes two hours, not two days.

Every line of code you don't write is a potential bug you don't have to fix.

Common pitfalls to avoid

Pitfall one: perfectionism. An MVP doesn't need to be beautiful. It needs to ship.

Pitfall two: premature scaling. You don't need a load balancer or a caching layer. 100 users run fine on Vercel's free plan.

Pitfall three: too many features. You'll be tempted to add just one more thing, then another. Don't. Ship the minimum and let users tell you what's missing.

Pitfall four: no deploy plan. Deploy daily. Small failures are cheap.

Pitfall five: forgetting to share the link. You built it: share it. Post in relevant communities and get feedback.