How to build a minimum viable product: a step-by-step guide

Executive overview

A minimum viable product (MVP) is the smallest thing you can build to validate a hypothesis with real customers — it does not have to be software. The core insight is that talking to customers is insufficient; you must put something in front of them to see how they actually behave. Rob Walling outlines three practical approaches — human automation, no-code, and full code — and argues that most founders overbuild and over-assume. The Zappos founding story illustrates how a billion-dollar company was validated with nothing more than borrowed store inventory and manual shipping.

What an MVP actually is

• An MVP is the minimum thing needed to test your next hypothesis with real customers.
• It is not a simplified version of your final product; the scope and shape may differ entirely.
• Its purpose is to determine whether the problem is worth solving and whether your approach is directionally correct.
• Customer interviews alone are unreliable — people accidentally lie or tell you what you want to hear.
• Mock-ups and surveys produce "messy data"; only a working solution reveals genuine behaviour.

When to build an MVP

• Build one when you suspect a problem exists but are not certain it is a desperate pain point.
• Build one even when you are confident — overconfidence is the default failure mode of founders.
• Use prior conversations, firsthand observation, or validated complaints as inputs, not as final proof.
• Everything should be treated as a hypothesis until real customer usage proves otherwise.

Approach 1: human automation (Wizard of Oz)

• The product appears automated to customers but is actually fulfilled by people behind the scenes.
• Example: selling curated sales leads to freelancers using virtual assistants crawling the web into a spreadsheet instead of building a scraper.
• The goal is not scalability or profitability — it is answering "will anyone pay for this?"
• Eliminates months of engineering before you know whether the market exists.

Approach 2: no-code

• Suitable for create/read/update/delete workflows, project management tools, and notification-heavy systems.
• Rob Walling's own podcast and YouTube production system was built entirely in Airtable by a non-developer.
• Key platforms: Bubble, Airtable, Zapier, and Make.
• No-code can produce robust tools quickly without writing a single line of custom code.

Approach 3: full code

• Appropriate when no-code cannot meet the minimum requirements of the hypothesis.
• Developer founders can use this path; non-technical founders face stacked odds.
• Hiring an agency or freelance developer is risky if you cannot evaluate technical quality.
• Common outcome for non-technical founders: six to twelve months of work, significant technical debt, and a rewrite.
• Recommended path for non-technical founders who want SaaS: find a developer co-founder.

What you can cut from an MVP

• Billing code — use Venmo, PayPal subscriptions, Stripe payment links, or manual invoicing.
• Refunds — handle manually.
• Password resets — not needed early on.
• Delete buttons — delete records manually from the database.
• The question to ask: what is the single core feature without which the MVP cannot exist?

How long should an MVP take?

• Simple AI wrappers or workflow tools: a weekend.
• More complex solutions: a few months of evenings and weekends.
• If you are exceeding a few hundred hours, you are likely overbuilding or tackling excessive complexity.
• Scope the MVP in writing and validate that scope with potential customers before building.

The Zappos example

• The founder wanted to test whether people would buy shoes online.
• He photographed inventory in local shoe stores and listed the photos online without purchasing any stock.
• When a customer ordered, he bought the shoes at full retail price and shipped them himself.
• No inventory investment, no e-commerce platform, no warehouse — pure human automation combined with no-code storefront.
• The company later sold to Amazon for $1.2 billion.
• The lesson: even a billion-dollar idea can be validated with a manual, low-cost MVP.