//pragmatic leaders

MVP Experiment

Reading time
6 min
Section
Minimum Viable Product - PLPM
6 min left0%
mvp experiment0%
6 min left
An MVP is not just a product. It is an ongoing process of learning and discovery through experiments.
Talvinder Singh, from a Pragmatic Leaders session on MVP experiments

Running an MVP experiment is not a one-time event. It is a step-by-step process that unfolds over multiple iterations to discover what your customers truly value. The MVP is not just a minimal product; it is a method to validate your assumptions with real users before investing heavily.

This page teaches you how to run an MVP experiment from start to finish, using examples grounded in Indian startups and practical product management.

Start by listening to the voice of the customer

The first step of any MVP experiment is to listen carefully to your customers. Their voice directs you to the actual problem you need to solve — not the problem you imagine or the one stakeholders describe.

As a PM, you must constantly ask:

  • Who will buy this product?
  • Why would they buy it?
  • How will they buy it?

For example, Uber’s founder noticed that people wanted to rent premium black cars from local taxi companies but were unable to do so because of the high costs. This insight came from observing customer pain points and unmet needs.

// scene:

Product discovery session at an early-stage startup

You (PM): “Who exactly is our customer? What problem are they trying to solve?”

Customer Researcher: “We heard from office-goers that timely rides are a huge pain point, especially during rush hours.”

You (PM): “Why do they prefer premium black cars? Is it status, comfort, or reliability?”

Customer Researcher: “Mostly reliability and safety. They want a ride they can trust, but the existing options are too expensive.”

This conversation shapes the problem statement and guides solution brainstorming.

// tension:

Understanding the customer's true needs is the foundation of the MVP experiment.

Identify and prioritize assumptions with user personas

Once you have a problem and brainstormed possible solutions, you have made many assumptions — often without realizing it.

Understanding your target audience deeply is crucial to making these assumptions accurate. User personas help by representing different customer segments with distinct needs and behaviors.

However, you cannot test all assumptions at once. You must prioritize by focusing on the riskiest assumptions first. These are assumptions that, if proven wrong, would be devastating to your product’s success.

For instance, Uber assumed that timeliness was the most important attribute for a customer segment like office goers. That was a riskiest assumption worth validating early.

// thread: #product-discussion — Prioritizing assumptions in the product team
Neha (PM)We have 15 assumptions about our users. Let's rank them by risk and impact.
Rahul (UX)Timeliness and safety seem to be the riskiest assumptions for early adopters.
Meera (Data)We can design experiments to test these two first.
YouGreat. Let's focus our MVP hypotheses around these.

Build testable hypotheses from assumptions

Developing a product is like a scientific experiment. You start with assumptions; then you build hypotheses that are measurable, actionable, and have clear outcomes.

Unlike vague assumptions, hypotheses must be structured so you can validate or invalidate them with data.

For example:

AssumptionHypothesis
Timeliness is important to office goersIf 80% of office-goer users receive a ride within 5 minutes, then 70% will rate the service as satisfactory

This hypothesis is measurable (80% rides within 5 minutes), actionable (we can improve dispatch algorithms), and has an outcome (70% satisfaction).

// scene:

MVP planning workshop

You (PM): “How do we turn assumptions into hypotheses?”

Data Analyst: “We define measurable success criteria and the expected user behavior.”

You (PM): “So for the timeliness assumption, can we say 70% satisfaction if rides arrive within 5 minutes?”

Data Analyst: “Yes, that’s a strong hypothesis to test.”

// tension:

Hypotheses must be testable to drive validated learning.

Set minimum success criteria and metrics

To measure your hypotheses, define minimum criteria for success. This is the break-even point that validates or invalidates your assumptions.

For example, if your hypothesis is about customer satisfaction, your minimum success criterion might be a satisfaction score above 70% in surveys during the MVP test period.

Metrics must be specific, quantifiable, and tied directly to your hypotheses. Common metrics include:

  • Customer satisfaction scores
  • Task completion rates
  • Time to first response
  • Retention rates
  • Net promoter score (NPS)
// exercise: · 10 min
Define your MVP success metrics
  1. List your current hypotheses.
  2. For each hypothesis, identify one or two quantifiable success metrics.
  3. Define the minimum threshold that would validate the hypothesis.
  4. Consider how you will collect this data during your MVP experiment.

MVP experiments test product viability, not just technical feasibility

The primary goal of MVP experiments is to test if the product is worth building, not just if it can be built technically.

You will learn more by launching a minimal solution to early adopters than by building a full-featured product without validation.

For example, Zappos started by manually photographing shoes at local stores and fulfilling orders by visiting those stores — no inventory, no advanced website. This MVP tested whether customers would buy shoes online before investing in technology or inventory.

Similarly, AirBnB’s early MVP was a simple website listing their own apartment during a conference, validating demand before building a full platform.

// thread: #mvp-experiments — Deciding MVP scope
Anjali (PM)Should we build the whole app or just a landing page for our MVP?
Vikram (Founder)Let's start with a landing page and manual order fulfillment to validate demand.
YouThat aligns with MVP principles — test riskiest assumptions with least effort.

Plan and execute your MVP experiment with data collection

After defining hypotheses and success criteria, you need to plan how you will collect the data to validate them.

A data collection plan answers:

  • What data is needed for each success criterion?
  • Where will the data come from?
  • How much data is required for statistical confidence?

For example, if customer satisfaction is a success criterion, you may decide to collect survey responses from at least 100 users over four weeks.

Once the plan is in place, launch the MVP to the right user segment and collect data diligently.

// scene:

Sprint planning meeting

You (PM): “We’ll collect satisfaction surveys weekly, monitor ride timings via logs, and track retention over 30 days.”

Data Engineer: “I'll set up dashboards for real-time monitoring.”

You (PM): “Great. Let’s run the MVP for six weeks and review the data every two weeks.”

// tension:

Executing MVP experiments requires discipline and clear data plans.

Collect feedback continuously and validate assumptions

As your MVP reaches the market, early adopters will begin giving feedback.

Surveys are the most popular way to collect structured feedback. Tools like Qualtrics, SurveyMonkey, and Google Forms are commonly used.

Other methods include live chat and customer calls, though the latter are often less scalable and less effective for statistical validation.

// thread: #customer-feedback — Using feedback to iterate
Meera (Customer Success)Survey responses show 65% satisfaction after two weeks.
YouBelow our 70% success criterion. Let's dig into qualitative feedback for pain points.
Rahul (UX)Users mention delays during peak hours and app crashes.
YouWe’ll prioritize reliability fixes in the next sprint.

Iterate the MVP cycle until assumptions are validated or invalidated

Product development is a continuous process. Each MVP cycle teaches you something new.

You will find some assumptions validated, some disproven, and some conditionally true.

Use these insights to refine your product, your hypotheses, and your experiments.

By the fourth or fifth MVP cycle, your experiments should become more predictable and your product more aligned with user needs.

// exercise: · 15 min
Plan your MVP iteration
  1. Review the data and feedback from your MVP experiment.
  2. Identify which hypotheses were validated and which were disproven.
  3. Prioritize the next set of riskiest assumptions or new questions.
  4. Design your next MVP iteration to test these.
  5. Set a timeline and data collection plan for the next experiment.

The MVP experiment timeline varies by scale

MVP experiments typically run between two and twelve weeks.

The duration depends on:

  • The scale of your user base
  • The sample size required for statistical confidence
  • The complexity of the product or feature

For example, a large company like Swiggy can run experiments on 1% of its millions of users and get fast results. An early-stage startup with few users may need longer to gather meaningful data.

Test yourself: MVP experiment planning

// learn the judgment

You are a PM at a seed-stage Bangalore startup building a hyperlocal grocery delivery app. You have identified that 'delivery speed' and 'product freshness' are the riskiest assumptions that determine customer retention. You have 500 early adopters signed up for your MVP test.

The call: How would you design your MVP experiment to validate these assumptions? What metrics would you track? How long would you run the experiment before concluding?

Your reasoning:

// practice

You are a PM at a seed-stage Bangalore startup building a hyperlocal grocery delivery app. You have identified that 'delivery speed' and 'product freshness' are the riskiest assumptions that determine customer retention. You have 500 early adopters signed up for your MVP test.

Your task: How would you design your MVP experiment to validate these assumptions? What metrics would you track? How long would you run the experiment before concluding?

your reasoning:

0 chars (min 80)

Where to go next

PL alumni now work at Flipkart, Google, Razorpay, PhonePe, Swiggy, Amazon, Microsoft, and 30+ other companies.