Summary of "Lean Analytics"

2 min read

Core Idea

  • Stop guessing; use data to kill entrepreneurial delusion—instincts are experiments, data is proof.
  • Pick ONE metric per stage that answers your riskiest question; everything you build must directly impact it.
  • Your business model determines your metrics—e-commerce, SaaS, mobile, media, and marketplaces each need different KPIs.

The Five Stages & Your OMTM (One Metric That Matters)

1. Empathy — Validate the problem is real

  • Run 15+ customer interviews; score for pain signals
  • Define OMTM: Can we prove customers have this problem?

2. Stickiness — Prove users come back

  • Focus on retention/engagement before scaling
  • Define OMTM: What % of users return daily/weekly?

3. Virality — Measure organic growth

  • Track viral coefficient and cycle time
  • Define OMTM: How many new users does each user bring?

4. Revenue — Prove unit economics work

  • Balance customer lifetime value vs. acquisition cost
  • Define OMTM: Does LTV exceed CAC by 3x+?

5. Scale — Grow efficiently across channels

  • Segment churn by channel to catch cannibalization
  • Define OMTM: Can we maintain growth metrics while expanding?

Avoid These Traps

  • Vanity metrics (total signups, page views, followers) stroke ego but don't drive decisions—use ratios and rates instead.
  • Asking users to rate features—force binary choices between alternatives to reveal true priorities.
  • Premature polish—kill features that don't improve OMTM; test with MVPs and proxies first.
  • Trying both efficiency AND differentiation—pick one strategy or get stuck in the middle.

What to Measure by Business Model

  • E-commerce: conversion rate, cart size, repeat purchase rate, customer lifetime value.
  • SaaS: churn, engagement, conversion to paid, monthly recurring revenue, customer acquisition cost.
  • Mobile apps: DAU/MAU, launch rate, ARPU, % paying users, time to first purchase.
  • Marketplaces: buyer/seller growth, inventory quality, search effectiveness, fraud prevention.

Run Experiments Like This

  • Define success before you start—draw a line in the sand quantitatively.
  • Combine qualitative + quantitative—interviews explain why, data proves what.
  • Use leading indicators (predict future) over lagging ones (explain past).
  • Prove causation, not just correlation—run controlled A/B tests to prove you can control outcomes.

Leadership & Scale

  • Three-Threes Model: 3 big assumptions (board), 3 actions/week (exec), 3 experiments/day (team) keeps everyone aligned.
  • Segment customers A/B/C: Fire or fix C-customers; double down on profitable B's; manage A's carefully to avoid churn.
  • Start with one secondary metric, optimize it publicly, use wins to justify data adoption company-wide.
  • Share methodology, not just results—transparency builds trust; failures teach more than wins.

Action Plan

  1. This week: Define your stage (Empathy/Stickiness/Virality/Revenue/Scale) and pick ONE metric that answers your riskiest question.
  2. Before building: Run 15+ customer interviews or 5-10 customer calls; validate pain or demand exists.
  3. Each sprint: Set a clear success threshold for your OMTM; kill features that don't move it; A/B test one variable at a time.
  4. Every decision: Ask "Does this improve our OMTM?"—if no, don't do it.
  5. Share results weekly: Show methodology + data to your team; let evidence, not ego, drive direction.
Copyright 2025, Ran DingPrivacyTerms
Summary of "Lean Analytics"