Principles of Product Management: Lean Software Development, MVP (Minimum Viable Product, Types: Concierge, Wizard of Oz, Piecemeal)

Lean Software Development, MVP (Minimum Viable Product, Types: Concierge, Wizard of Oz, Piecemeal)

Lean Software Development & MVP (Minimum Viable Product) – Study Guide

What This Is

Lean Software Development (LSD) is a methodology that applies lean manufacturing principles (eliminate waste, amplify learning, decide late, deliver fast) to software. The MVP (Minimum Viable Product) is its most famous tool—a version of a product with just enough features to validate core assumptions with real users, while minimizing cost and time. It’s not a "smaller" product; it’s a strategic experiment to test hypotheses (e.g., "Will users pay for this feature?").

Why it matters: In a world where 90% of startups fail (CB Insights), LSD/MVP reduces risk by forcing teams to learn fast before over-investing. Example: Dropbox’s MVP wasn’t a half-built app—it was a 3-minute video demonstrating the product concept. The waitlist grew from 5,000 to 75,000 overnight, proving demand before writing a line of code.

Key Terms & Frameworks

• Lean Software Development (LSD) Principles:
• Eliminate waste (e.g., unused features, excessive documentation).
• Amplify learning (use MVPs, A/B tests, user feedback).
• Decide as late as possible (avoid premature optimization).
• Deliver as fast as possible (short cycles, continuous deployment).
• Empower the team (trust engineers/designers to solve problems).
• Build integrity in (focus on UX, not just features).

• See the whole (avoid local optimizations that hurt the system).

• MVP (Minimum Viable Product): A product with the minimum features needed to test a core hypothesis (e.g., "Will users share files if it’s 1-click?"). Not a "v1" or "beta"—it’s an experiment.

• Types of MVPs:

• Concierge MVP: Manually deliver the value (e.g., a PM personally onboards users to test a "smart" feature before automating it). Example: Zappos started by taking photos of shoes in stores to test demand.
• Wizard of Oz MVP: Fake automation (e.g., humans behind the scenes power a "AI" feature). Example: Airbnb’s early "professional photography" service was just the founders taking photos themselves.

• Piecemeal MVP: Stitch together existing tools to simulate the product (e.g., using Typeform + Stripe + Zapier to test a subscription service). Example: Groupon started as a WordPress blog with PDF coupons.

• Build-Measure-Learn (BML) Loop (Eric Ries, The Lean Startup):

• Build (MVP to test a hypothesis).
• Measure (track leading/lagging metrics).
• Learn (validate/invalidate the hypothesis).

• Pivot or persevere (adjust or double down).

• ICE Score (Sean Ellis): Impact × Confidence × Ease – Prioritize experiments.

• Impact: How much this moves the needle (1–10).
• Confidence: How sure you are (1–10, based on data/user feedback).

• Ease: How hard it is to build (1–10, 10 = easiest).

• Pirate Metrics (AARRR): Acquisition-Activation-Retention-Revenue-Referral – Track these for your MVP to identify leaks in the funnel.

• Leading vs. Lagging Indicators:

• Leading: Predict future success (e.g., % of users who complete onboarding).

• Lagging: Reflect past success (e.g., revenue, churn). MVP focus: leading indicators.

• Vanity Metrics vs. Actionable Metrics:

• Vanity: "10,000 signups!" (no insight into behavior).

• Actionable: "30% of signups complete onboarding" (tells you what to fix).

• Pivot Types (Eric Ries):

• Zoom-in: One feature becomes the whole product (e.g., Flickr started as a game, pivoted to photo-sharing).
• Zoom-out: Expand scope (e.g., Twitter started as a podcasting platform).
• Customer segment: Same product, new audience (e.g., Slack pivoted from a gaming company to enterprise chat).
• Technology: Same problem, new tech (e.g., Netflix from DVDs to streaming).

Step-by-Step Process Flow

1. Define the Core Hypothesis
2. Action: Write a testable hypothesis (e.g., "We believe [user segment] will [do X action] because [reason], and we’ll know this is true when we see [metric].").

3. Example: "We believe freelancers will pay $10/month for a tool that auto-generates invoices because they hate manual work, and we’ll know this is true when 20% of trial users convert to paid."

4. Choose the Right MVP Type

5. Action: Pick the cheapest way to test the hypothesis:
   • Concierge: If the "smart" part is hard to build (e.g., AI recommendations).
   • Wizard of Oz: If automation is complex (e.g., chatbots).
   • Piecemeal: If you can stitch together existing tools (e.g., no-code MVP).

6. Example: To test a "smart resume builder," start with a Concierge MVP where a PM manually edits resumes for users.

7. Build the MVP (Fast & Cheap)

8. Action:
   • Limit scope to one core feature (e.g., "auto-fill work history" vs. "full resume builder").
   • Use no-code tools (Bubble, Zapier, Airtable) or manual processes (e.g., Google Forms + Sheets).
   • Avoid: Polishing UI, adding "nice-to-haves," or over-engineering.

9. Example: Buffer’s MVP was a 2-page website: a landing page to gauge interest + a pricing page to test willingness to pay.

10. Measure & Learn

11. Action:
    • Track leading indicators (e.g., % of users who complete the core action).
    • Run qualitative interviews (e.g., "Why didn’t you finish onboarding?").
    • Use A/B tests if possible (e.g., "Does a video demo increase conversions?").

12. Example: Twitter’s MVP (a SMS-based service) tracked "tweets per user" to validate engagement.

13. Decide: Pivot or Persevere

14. Action:
    • If metrics hit targets-persevere (build v2).
    • If metrics miss-pivot (change hypothesis, audience, or feature).
    • Rule of thumb: If <10% of users complete the core action, pivot.
15. Example: Instagram pivoted from a location-based app (Burbn) to photo-sharing after users ignored most features.

Common Mistakes

• Mistake: Building a "smaller version" of the full product (e.g., a "lite" app with 50% of features).

• Correction: An MVP tests one core hypothesis (e.g., "Will users pay for this?"). Strip everything else. Why? Overbuilding wastes time and obscures learning.

• Mistake: Ignoring qualitative feedback (e.g., only tracking metrics like "signups").

• Correction: Pair metrics with user interviews (e.g., "Why did you abandon the cart?"). Why? Metrics tell you what happened; interviews tell you why.

• Mistake: Launching an MVP without a clear success metric (e.g., "Let’s see what happens").

• Correction: Define one key metric before launch (e.g., "20% of users complete onboarding"). Why? Without a target, you can’t validate the hypothesis.

• Mistake: Confusing "MVP" with "MMP" (Minimum Marketable Product).

• Correction:

  • MVP: Tests a hypothesis (e.g., "Will users share files?").
  • MMP: The smallest product that can be sold (e.g., Dropbox’s first public release). Why? MMPs come after validating the MVP.

• Mistake: Over-optimizing for "perfect" data (e.g., waiting for 10,000 users to decide).

• Correction: Use small sample sizes (e.g., 50–100 users) for qualitative feedback. Why? Speed > perfection in early-stage learning.

PM Interview / Practical Insights

1. Tricky Distinction: MVP vs. Prototype vs. Proof of Concept (PoC)
2. MVP: Tests a business hypothesis (e.g., "Will users pay?").
3. Prototype: Tests design/UX (e.g., "Is the onboarding flow intuitive?").
4. PoC: Tests technical feasibility (e.g., "Can we build this feature?").

5. Interview trap: "How would you test if users want a voice-based search feature?"-Answer with an MVP (e.g., Wizard of Oz), not a prototype.

6. Stakeholder Pushback: "This MVP is too ugly/basic!"

7. How to respond:

   • "The goal isn’t to impress users—it’s to learn as fast as possible. If we polish it now, we risk wasting months on the wrong thing."
   • "Let’s test this with 50 users. If they hate it, we’ll iterate. If they love it, we’ll invest in scaling."

8. Interview Question: "How would you decide between a Concierge vs. Wizard of Oz MVP?"

9. Answer:

   • Concierge: Use when the value is manual (e.g., personalized recommendations).
   • Wizard of Oz: Use when the automation is the risk (e.g., "Can we build an AI that does X?").
   • Example: For a "smart expense categorizer," start with Concierge (PM manually categorizes) to test if users care. Later, use Wizard of Oz (humans pretend to be AI) to test if automation works.

10. Leading vs. Lagging Indicators in MVPs

11. Interview trap: "What metrics would you track for an MVP?"-Don’t say "revenue" (lagging). Say leading metrics like "% of users who complete the core action" or "time to first value."

Quick Check Questions

1. Scenario: Your team wants to build a "smart" calendar that auto-schedules meetings based on user preferences. The engineers say it’ll take 6 months. How would you test this idea this week?

2. Answer: Concierge MVP – Manually schedule meetings for 10 users and track if they use the feature. Why? Tests demand before investing in automation.

3. Scenario: Your MVP’s conversion rate is 5% (target was 20%). The team wants to add more features to "improve" it. What do you do?

4. Answer: Pivot or dig deeper – Interview users to find why they’re not converting (e.g., "The onboarding is confusing"). Why? Adding features without understanding the root cause wastes time.

5. Scenario: A stakeholder says, "This MVP is too simple—users will think we’re unprofessional." How do you respond?

6. Answer: "Let’s test it with 50 users. If they complain about simplicity, we’ll iterate. If they don’t, we’ll know we’re on the right track." Why? Data > opinions.

Last-Minute Cram Sheet

1. MVP = Experiment, not a product. Goal: Learn, not impress.
2. Concierge MVP: Manual delivery (e.g., PM does the work).
3. Wizard of Oz MVP: Fake automation (e.g., humans pretend to be AI).
4. Piecemeal MVP: Stitch together existing tools (e.g., no-code).
5. Build-Measure-Learn (BML): The core loop of Lean Startup.
6. ICE Score: Impact × Confidence × Ease – prioritize experiments.
7. Leading indicators > lagging indicators for MVPs (e.g., % onboarding completion vs. revenue).
8. Pivot types: Zoom-in, zoom-out, customer segment, technology.
9. MVP-MMP – MMP is the smallest sellable product.
10. Vanity metrics (e.g., "10K signups")-actionable metrics (e.g., "30% retention").