The most crowded corridor of the startup graveyard is "products nobody wanted": 42% of startups fail for lack of a market. MVP thinking exists precisely against that risk — and is widely misunderstood.
What an MVP is — and is not
An MVP is the smallest version of your product that tests the core value proposition. Not a buggy half-product: a product that does little, but does it well.
Step 1: A one-sentence hypothesis (week 1)
Fill in: "[Audience] has [problem]; if we offer [solution], they will [measurable behavior]." If you cannot write this sentence, development should not start.
Step 2: Ruthless scope cutting (weeks 1–2)
Sort features into three buckets: Core (without it the hypothesis cannot be tested — usually 3–5 features), Later, and Maybe never. Painful truth: 70% of your initial list does not belong in the first two. Excel instead of an admin panel, hand-made invoices instead of automation — doing things that don't scale is the MVP superpower.
Step 3: Technology choice (week 2)
One criterion: iteration speed. Microservices and Kubernetes are MVP enemies. Our typical stack: Next.js/React + Node.js + PostgreSQL. One server carries your first 10,000 users — reaching that "problem" is success. (Undecided on the database? PostgreSQL vs MongoDB.)
Step 4: The 6-week build sprint (weeks 3–8)
- Weeks 3–4: skeleton of the core flow, end to end
- Weeks 5–6: deepen core features + payments/analytics
- Week 7: closed beta with 10–20 real users
- Week 8: critical feedback + launch
Step 5: The learning loop
The MVP's output is learning, not product. Track three metrics: activation (how many signups reach the value moment), retention (do they come back after a week — see our retention guide) and referral. If these don't move, more features are not the answer — revisit the hypothesis.
Budget reality
A professionally built MVP typically runs $15–50k depending on scope. Read it as the cost of the market test — always cheaper than fully building the wrong product for 18 months. Have an idea? Book a free MVP scoping call.