Concept · Innovation

Minimum viable product

A minimum viable product is the smallest responsible experiment that can test a consequential product or business hypothesis with useful evidence.

A focused test converts an assumption into evidence.

A minimum viable product is a deliberately bounded way to obtain evidence for a consequential product or business decision. Its minimum depends on the hypothesis and what the participant must experience to answer it. It is not automatically a smaller feature list or a cheaper version of the intended product.

In one sentence

A minimum viable product is the smallest experiment or product form that can generate credible learning about a specified customer or business hypothesis.

The decision is which uncertainty to test before the next commitment. A prototype can test comprehension, a manual service can test useful value and a limited live product can test operating behavior. Those methods answer different questions.

Validity and responsible delivery constrain the minimum. A shortcut that removes the task’s crucial condition may produce cheap but misleading evidence. A test must preserve the conditions necessary for the inference and the participant’s reasonable expectations.

MVP forms

Choose the lowest-cost method that can answer the riskiest question without misleading or harming participants.

Concept or prototype

Test comprehension, workflow, or usability before building production infrastructure.

Manual or concierge service

Deliver the outcome by hand to learn whether the job matters and what work it requires.

Limited product release

Ship a constrained but safe feature to a defined cohort and measure real use and retention.

A continuum, not a switch

MVPs range from low-fidelity artifacts to limited live products. The right minimum is determined by the decision, evidence quality, and participant risk—not by how small the feature list is.

LowLarge build before learningHighFocused experiment before commitment

“Viable means useful for the test and responsible for the user.”

Why it matters

Steve Blank’s account distinguishes a customer-value question from building the technology imagined to deliver it. The example supports matching the test to the decision. The anecdote does not establish a universally successful method.

The mechanism is an observation capable of changing the next action. State the question, the meaningful customer event and the limit of what the result can show. A favorable usability response cannot supply willingness to pay or the cost of reliable service.

Interest can reflect novelty, a demonstration or personal assistance. A negative result can reflect poor execution rather than absence of need. Inspect whether the test actually exposed the intended condition before choosing a broader commitment.

Real-world examples

The same concept shows up in different ways across industries.

DropboxEntry is not a published activation experiment

→

Dropbox describes a free entry route and claims quick setup. The registration supplies no distribution of time to useful work and no comparison of onboarding variants. An MVP lesson should not invent those results from the signup design.

NetflixA bounded comparison has an explicit target

→

Netflix’s artwork account supplies an actual test-and-rollout sequence: a contextual policy was evaluated against an unpersonalized policy, with attention to engagement quality. The decision concerned a selection method inside an existing service, not the launch of a new minimal product. It is useful here because it shows how the evidence requirement follows the uncertainty. A demonstration can establish that the selection method operates; it cannot by itself establish that viewers receive better recommendations. Offline replay can help screen a policy under logged conditions, while the reported online comparison addresses behavior under the new choice. The account does not disclose a complete numerical causal report or prove a retention effect. For a hypothetical new product, choose the smallest test that can change the relevant decision: a manual service may test a customer task, while a functional prototype may be necessary to test technical reliability. Those alternatives differ because they answer different questions. Minimal scope is a means of learning, not evidence that every commercially important outcome can be inferred from a brief demonstration.

When it breaks

A hypothetical prototype is easy to use while the live service depends on unreliable external data. The prototype answers an interface question and leaves operating feasibility open. Name that boundary before treating it as product validation.

A manual service can reveal valuable work while hiding the labor needed to scale. Preserve what was learned and investigate the next constraint. Cheap evidence is useful only when its scope is clear enough to guide a decision.

Key takeaways

01
Name the riskiest hypothesis and decision before building.
02
Use the least costly test that can produce credible evidence.
03
Preserve the conditions and participant commitments needed for a valid, responsible test.

Sources

An MVP Is Not a Cheaper Product, It’s about Smart Learning · Steve Blank. Keep Your Eyes on the Prize; Lessons Learned
Dropbox registration statement on Form S-1 · Dropbox / SEC. Our Business Model: signup, acquisition and paid conversion paragraphs; paying-user definition printed p. 13; registration/conversion risk p. 15; referral and enterprise selling risk; Sales and Marketing
Artwork Personalization · Netflix TechBlog. Contextual bandit approach, model training and performance evaluation, December 2017