MVP development creates the smallest useful product that can test an important business assumption with real users. Build one when you need to see whether people complete a real workflow, return or pay. If a guided prototype or a manual service can answer the question first, use that cheaper test before committing to a software build.
TL;DR
- Build an MVP when real use or payment must test your idea; use a prototype when a guided test can answer the question.
- Keep one end-to-end workflow, essential quality and a way to measure it. Defer features that do not change the first decision.
- A focused web MVP may use an illustrative 8–16-week plan; team availability and integration access can change it.
- One supplier estimates $80,000–$160,000 for a web portal MVP; that is its estimate, not a market average or a Hapy quote.
- A hypothetical $80,000 build plus $2,000 monthly operation costs $104,000 over the build and first 12 months after launch.
- Set a pilot decision before launch: count completed outcomes and payment or repeat use, then decide whether to continue, change or stop.
A minimum viable product still needs enough quality for someone to get the promised result. A broken payment step or an unclear status can obscure whether the idea works. The goal is to limit the scope while keeping the first experience trustworthy.
What should an MVP prove?
Choose the business decision before the feature list. A founder might need to know whether customers will pay for a service, whether a workflow saves enough time, or whether people return without a reminder. The first release should make that decision possible.
Take a hypothetical invoice product for independent consultants. The assumption is that consultants will use a simple invoice and payment-status workflow instead of following up manually. The first version lets a consultant create an invoice, send a payment link and see its status; the founder can handle unusual corrections manually. An accounting dashboard, subscription tiers and automated tax reports can wait because the first test does not require them.
| What to include | Why it matters in the invoice example | What can wait |
|---|---|---|
| One core user journey | Create, send and check an invoice | A second user type with a separate workflow |
| Essential trust and recovery | Access control, clear payment status and a way to correct an error | Advanced automation of rare exceptions |
| Measurement and feedback | Count sent invoices, successful payments and repeat use | A complex analytics dashboard |
| A named operating owner | Someone handles payment questions and failed notifications | A fully staffed support function before demand is known |
That example is a scope decision, not a recommendation to omit controls required by a real payment or privacy arrangement. The MVP checklist helps turn the selected workflow into acceptance criteria. For the common scope and launch failures that can weaken the test, see MVP development challenges.
Prototype, manual service, or working MVP?
Use the least expensive test that can produce evidence for the decision. A prototype can show a sequence of screens to test whether people understand an offer. A manual or “concierge” service can test whether they want the outcome while people perform the work behind the scenes. A working MVP is useful when actual use, repeat behavior or payment is the question.
| Approach | Best question to answer | What it does not prove by itself |
|---|---|---|
| Prototype | Can a user understand and navigate the proposed experience? | That the service can be delivered reliably or that users will return |
| Manual service | Will a customer request or pay for this outcome? | That a proposed automated system works at scale |
| Working MVP | Can users obtain the outcome and repeat or pay for it? | That a larger product or business model will succeed |
Y Combinator’s MVP planning guidance also frames the first release around a specific problem and a narrow feature set. See proof of concept, prototype and MVP when the main uncertainty is technical feasibility rather than customer behavior.
How do you plan the development work?
Start with a pilot group and the result it needs. Map the shortest path from the first action to that result, including failure and recovery. Then agree what the team will build, what can be done manually, what will be measured and who makes scope decisions. The step-by-step build guide covers the delivery work in more detail.
For a narrow web MVP, an illustrative eight-to-sixteen-week plan can cover one workflow, two engineers, part-time design and quality assurance, one established integration and a product owner who answers questions within two working days. The team can overlap design, development and testing. This is a planning heuristic, not a measured industry average or a Hapy delivery promise. Native apps, untested APIs, data migration, procurement and regulated workflows can change the schedule.
ScienceSoft’s MVP guide, checked in November 2026, describes code-based development taking two to six months in its process. That supplier’s range covers different scopes; it does not validate a deadline for your project. A proposal should show the team, dependencies, launch checks and the time reserved for a pilot, not just the coding period.
What should a founder budget?
Budget for the first usable release and the time needed to operate and learn from it. ScienceSoft estimates $80,000–$160,000 for a web portal MVP and $40,000–$300,000 or more across its broader MVP examples. It states that these are estimates, not its official project prices. They describe one supplier’s view and are not a market average or a Hapy quote. A simpler manual or no-code test may need a very different budget.
Here is a hypothetical comparison for one portal. Assume an $80,000 build and $2,000 per month after launch: $800 for hosting and software, $800 for support and fixes, and $400 for other operating services. The build plus twelve months of operation totals $104,000: $80,000 + (12 × $2,000). The figures are planning assumptions, not quoted prices. Founder salary, sales, marketing, transaction fees and taxes are outside this example.
Ask a delivery partner to separate discovery, design, engineering, quality assurance, launch and support. Account for the work your own team must do to supply content, make decisions and test the result. Use the MVP cost guide for a more detailed first-year budget and the partner-vetting guide to compare proposals.
How do you know whether the MVP worked?
Set the decision rule before users arrive. For the hypothetical invoice pilot, a founder might invite 40 consultants and decide to continue only if at least 12 send an invoice and at least 6 return to send another within 30 days. If 12 send one but nobody returns, the next step is to investigate the repeat-use barrier, not automatically build the deferred dashboard. These numbers are example thresholds chosen for the decision, not industry benchmarks.
Track the denominator as well as the result: 12 of 40 invited consultants completed the first action, or 30%. Record failures and feedback beside that number so a broken invitation or payment flow is not mistaken for lack of demand. Decide who will review the evidence and when the pilot ends.
Who should build the first version?
Use an existing team when it has product, engineering and testing capacity for the pilot. An external team is useful when the founder needs those skills for a defined release and can provide timely product decisions. A hybrid can work when an internal owner sets priorities and an external team supplies missing delivery capacity. Compare the responsibilities and handover in agency versus in-house MVP development.
Whichever model you choose, agree ownership of the code, designs, accounts, data and operating documentation before work begins. Hapy offers MVP development services; compare its scope, evidence and terms with the same criteria you apply to any other partner.



