To build an MVP, start with a specific customer and an important problem—not a list of features. Identify the riskiest assumption behind your idea, then run the least costly credible experiment that can test it. That experiment might be a landing page, mockup, or manually delivered service rather than a software release. Decide in advance what evidence would change your next step, observe what customers do as well as what they say, and use the result to continue, revise, or test again.
What an MVP is—and what it is not
A minimum viable product (MVP) is a way to learn about a value proposition or business model with minimal effort. It is not automatically a shrunken version of the final product. Its job is to test a consequential assumption: for example, whether a particular group has a problem, wants the proposed outcome, can be served feasibly, or will pay enough for the business to work.
Eric Ries, quoted in Strategyzer’s MVP guidance, describes an MVP as the fastest way to get through the Build-Measure-Learn feedback loop with minimum effort—not necessarily the smallest product imaginable. In practical terms, build only what is needed to produce useful evidence for the question you are asking.
How to build an MVP: a practical sequence
1. Specify the customer and problem
Describe who experiences the problem, when it occurs, and what outcome that person needs. Be precise enough that you can recruit the right people and recognize relevant evidence. A broad audience such as “small businesses” is rarely specific enough to guide a useful test; consider the role, context, or situation that makes the problem acute.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →In discovery conversations, ask about the customer’s jobs, pains, and desired gains rather than introducing your solution first. Ask for examples of what they do now and what makes the situation difficult. Interviews can reveal context, language, and mistaken assumptions; they do not by themselves establish that people will act or buy.
2. List and rank the risky assumptions
Write down the assumptions that must be true for the idea to work. Sort them into three categories:
- Desirability: Will the intended customers want, choose, or use this?
- Feasibility: Can the team deliver it technically and operationally?
- Viability: Can the business create value sustainably relative to its costs?
Choose the assumption whose failure would most change your plan. Testing a low-impact detail first can create the appearance of progress while leaving the central risk untouched. Strategyzer’s evidence framework also emphasizes checking whether evidence comes from users, decision-makers, or both; the person who uses a product may not be the person who chooses or pays for it.
Rank #2
3. Choose the lightest credible experiment
Work backward from the learning goal. First decide what observation would answer the question, then choose a measure that captures it, and only then create the artifact or service needed to produce that observation. A proxy is often faster and cheaper than building the intended product, as long as it can credibly test the assumption.
Possible experiments include:
- A value-proposition data sheet, brochure, or storyboard to test whether the problem and proposed outcome are clear and relevant.
- A landing page with a clear call to action to observe whether people take a next step.
- A mock product box or explanatory video to test reactions to a product concept before implementation.
- A basic learning prototype when the question depends on interaction or functionality.
- A Wizard-of-Oz or concierge MVP: customers see a product-like front end, while people perform some or all of the work manually behind it.
These are not rungs on a mandatory ladder. Pick the option that can answer the current question with enough fidelity to make the result interpretable. If a mockup cannot answer whether a core interaction is usable, a functional prototype may be necessary. If the question is whether a problem matters, building that interaction may be unnecessary.
4. Set the hypothesis and decision rule before the test
Record the hypothesis, target customer segment, action or outcome you will observe, and the threshold that would count as support, rejection, or uncertainty. Include who will be recruited and any constraints that affect the result. A threshold prevents a team from redefining success after seeing the outcome.
Rank #3
Evidence differs in strength. An interview or survey captures what someone says; a reaction to an artifact gives a more concrete response; a signup or other action shows a step taken; and a purchase or presale can indicate a more meaningful commitment. Strategyzer’s September 2, 2026 article places what people say at levels 1–2 and what people do at levels 3–5 on its evidence scale. That is Strategyzer’s framework, not a universal scientific standard.
5. Run the test, interpret the result, and choose a next step
Compare the observed result with the rule you set. If it supports the assumption, decide whether the evidence is strong enough for the next investment or whether another test is warranted. If it contradicts the assumption, consider changing the customer, problem, value proposition, or business model. If the result is ambiguous, revise the test or clarify the assumption instead of calling the idea validated.
A negative result is useful when the experiment tested the intended question. If few people respond to a landing page, for instance, the result may reflect weak demand, but it may also reflect confusing wording or poor recruitment. A test should be designed so that a disappointing outcome helps distinguish among plausible explanations.
Choose an experiment by the evidence it can produce
When multiple experiments seem possible, compare them against the question rather than defaulting to the one that looks most like a product.
| Experiment | Useful when | Evidence and trade-off |
|---|---|---|
| Interview or survey | You need to understand context, customer language, jobs, pains, or stated preferences. | Relatively light to conduct, but captures what people say rather than proving they will act or pay. |
| Storyboard, brochure, or mockup | You need reactions to a value proposition or concept before building it. | More concrete than a verbal description, but does not establish that a working product is usable. |
| Landing page with a call to action | You want to observe whether a defined audience takes a next step. | Captures an action, but results can be hard to interpret if the audience, message, or recruitment is a poor match. |
| Learning prototype | The assumption depends on interaction, usability, or technical behavior. | Can test functionality, but requires more effort than a simple artifact; build only enough to test the question. |
| Wizard-of-Oz or concierge service | You need to test the customer experience or demand before automating delivery. | Lets a product-like experience be delivered manually, but does not by itself prove that the operation can scale. |
For any option, ask whether it tests the important assumption, whether the outcome is an opinion, reaction, action, or commitment, how much time and effort it requires, and whether a failure would be interpretable. Make sure the people involved match the customer and buyer roles relevant to the decision.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Use interviews to discover assumptions, not manufacture validation
Interviews are most useful for learning how a problem appears in a customer’s life and how that person currently handles it. Ask about concrete past experiences rather than inviting a prediction about a hypothetical product. Visual artifacts or structured prioritization can make responses easier to compare, but the interview count alone does not establish demand.
A Strategyzer account of a Canadian consumer packaged-goods team’s customer-profile interviews illustrates why assumptions need testing. In that team’s 2026 example, 20 interviews led interviewees to delete at least once 91% of the team’s assumed customer-job statements because they did not apply; 55% of assumed pains were not true for those customers; and 75% of interviewees commented on at least one gain statement, refining or correcting the team’s understanding. These are case-specific findings, not expected rates or benchmarks for other teams.
The practical lesson is to treat a customer profile as a set of hypotheses to refine, not as a description confirmed by a few agreeable conversations. As Strategyzer’s Programme Design Lead Kurt Bostelaar puts it, “The more unproven an idea is, the more a customer-first focus is needed.”
Adapt the process to the stakes and constraints
There is no universal MVP budget, duration, sample size, or feature count established by the methods described here. The right scope depends on the customer, the assumption, the evidence needed, and the cost of being wrong. State your segment, threshold, and constraints so that a result is understood in context rather than treated as a general rule.
For regulated, safety-critical, or capital-intensive products, a low-cost proxy may help answer some questions but cannot replace domain-specific evidence or safeguards. The appropriate experiment depends on the product and the relevant constraints; no single MVP recipe covers those cases.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Further guidance
Strategyzer’s MVP article points to the book Value Proposition Design for deeper guidance. Strategyzer also describes an Understanding customers course covering discovery interviews, assumption mapping, and experiment selection. Its page displayed USD $400 when reviewed; price and availability may change.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
