MVP
Build the smallest useful test of the risky assumption. · Product Judgment · Lesson 09 · 3 min
MVP · 3 min
Situation
A smaller product can still be the wrong test.
You want to help small retailers reorder stock automatically. The team removes advanced reporting, themes, and integrations from the plan and calls the remainder an MVP.
But the central risk is whether shop owners trust the recommendations enough to act. A stripped-down interface does not necessarily test that trust.
Mental model
Start with the learning decision.
Name the risky assumption, the evidence needed, and the next investment that depends on it. Then choose the smallest useful experience that can produce that evidence.
For reordering, a weekly recommendation reviewed with five stores may reveal more than months spent building a self-service system.
Example
Manual behind the scenes, useful in front.
Have an analyst prepare reorder suggestions from data customers agree to share. Let owners approve, edit, or reject each suggestion. Record reasons and actual follow-through.
Be transparent about the service. The test can validate usefulness and trust; it does not yet validate automated accuracy, self-service acquisition, or scalable unit economics.
Failure case
Cutting the part that creates value.
If the test requires owners to manually retype hundreds of stock levels, non-use may reflect setup burden rather than lack of demand for recommendations.
A minimum experience still needs enough reliability and usability for the intended test. “We shipped something small” is not a learning outcome.
PM question
What happens after the result?
Before the pilot, decide what would justify automation, a narrower segment, or stopping. For example, study whether owners repeatedly act on recommendations and why they override them.
Do not treat five pilot users as proof of a whole market. Use the result to choose the next test, with the limits of the evidence stated.
Remember this
Minimum is relative to the assumption.
An MVP is a useful test with a decision attached. It can be manual, narrow, or temporary; it should not be carelessly unreliable.