lean-analytics

---
name: lean-analytics
description: Use when choosing startup metrics, auditing a dashboard for vanity metrics, or setting the One Metric That Matters. Applies Lean Analytics: metric trees by model, five stages, benchmarks.
category: Business Knowledge
version: 1.0.0
tools: []
---

# Lean Analytics

A data discipline for startups distilled from Alistair Croll and Benjamin Yoskovitz's *Lean Analytics*: separate metrics that change decisions from numbers that merely flatter, then point the whole company at the One Metric That Matters (OMTM) for your business model and stage.

## Core Principle

Focus on the one metric that matters right now -- everything else is noise that feels like progress. Startups die from lack of focus more than lack of data. A metric earns attention only if it changes what you do next.

## Scoring

**Goal: 10/10.**
- **9-10:** one OMTM matched to model and stage, a paired counter-metric, a line in the sand with a pre-committed miss response, cohorted and segmented data.
- **7-8:** mostly actionable ratios and a plausible OMTM, but no explicit target, weak cohorting, or too many "key" metrics.
- **5-6:** actionable and vanity metrics mixed; dashboard rarely changes a decision; model and stage never named.
- **3-4:** vanity metrics dominate -- totals, cumulative charts, blended averages; metrics copied from other companies.
- **0-2:** no instrumentation, or numbers chosen to impress investors rather than drive decisions.

## 1. Good Metrics vs Vanity Metrics
A good metric is comparative (vs. last week, vs. another cohort), understandable, a ratio or rate (not an ever-growing total), and behavior-changing -- if a number won't change what you do, stop measuring it. The output of analytics is decisions, not data. Work the lens pairs: qualitative vs quantitative, exploratory vs reporting, leading vs lagging, correlated vs causal -- correlation finds the lever, only a randomized experiment proves it. Cohorts (by signup month) and segments (by channel, plan, geography) keep comparisons honest; blended averages hide decay and mask whether one segment is soaring while another collapses. Averages lie under skew -- read medians and percentiles. A cumulative up-and-to-the-right chart is the single most reliable vanity tell. Application: rewrite every dashboard total as a ratio (total signups -> % of visitors activating within 7 days); report cohorted retention, not cumulative user curves.

## 2. The One Metric That Matters (OMTM)
At any moment there is one number that tells you whether the current riskiest assumption is working. Pick it, display it everywhere, and let it drive every experiment until you graduate to the next stage. It rotates -- it's the metric that matters *now*, not forever. Pair it with a counter-metric so it can't be gamed (activation speed paired with 30-day retention; sales velocity paired with refund rate). A line in the sand has three parts: a target number, a date, and a pre-committed answer to "what do we do if we miss?" -- "good enough" is decided in advance, not discovered after. Collect many metrics, but *watch* one; the rest live in drill-down reports. Ethical boundary: the line in the sand disciplines the company's bets, not individuals -- turning the OMTM into personal quotas invites gaming.

## 3. Metrics by Business Model
Your business model dictates which metrics exist and matter. Six archetypes, each with its own metric tree: **E-commerce** (conversion rate, average order value, repurchase rate -- annual repurchase under ~40% means acquisition mode, over ~60% loyalty mode). **SaaS** (MRR, churn, LTV:CAC, expansion, time-to-value). **Free mobile app** (downloads -> DAU/MAU, % paying, ARPDAU vs ARPPU -- whales skew every average). **Media** (audience, engaged time not raw pageviews, CTR, RPM). **UGC** (engagement funnel: visitor -> voyeur -> commenter -> creator, plus content-per-user and spam rate). **Marketplace** (liquidity: listings, fill/sell-through rate, time-to-transaction, take rate, buyer/seller ratio -- GMV is vanity until multiplied by take rate). Hybrid businesses must pick ONE primary model to own the OMTM; the secondary model contributes counter-metrics only.

## 4. Metrics by Stage: The Five Lean Analytics Stages
Startups move through five stages, each with a gate; working on a later stage's metric before passing the current gate is the canonical startup mistake. **Empathy:** have 15+ problem interviews shown a painful, frequent, paid-for problem? **Stickiness:** do people use it repeatedly on their own -- track retention cohorts, don't pour users into a leaky bucket. **Virality:** do users bring users -- track viral coefficient AND cycle time (shortening the cycle often grows faster than raising the coefficient). **Revenue:** does a dollar in return more than a dollar out, soon enough -- revenue per customer, CAC payback, gross margin. **Scale:** channels, partners, new markets -- metrics shift from product risk to ecosystem and operations. Gates are evidence, not time: a flattening retention curve exits Stickiness; positive unit economics within payback tolerance exits Revenue.

## 5. Baselines and Lines in the Sand
A metric without a target is trivia. Use published baselines as heuristics, not laws, then draw your own line in the sand. Reference points: early-SaaS monthly churn viable ceiling ~5%, healthy companies push toward ~2% or lower; DAU/MAU around 20%+ signals real engagement in habitual/social apps; casual mobile apps average roughly 14% day-30 retention; e-commerce typically converts ~1-3% of visitors; landing pages on good paid traffic convert low single digits, 25-30% is exceptional, not a planning number; a viral coefficient above 1 is rare and fleeting; no benchmark for your case -- measure your own baseline and watch the derivative, since 5% weekly improvement compounds into category-leading numbers. Benchmarks shift by market, channel, price point, and era -- always re-derive against your own cohorts.

## Common Mistakes

| Mistake | Why It Fails | Fix |
|---------|-------------|-----|
| A dashboard with 40 metrics | Diffuses focus; nobody owns anything | One OMTM big, 4-6 supporting metrics, archive the rest |
| Celebrating cumulative charts | Totals can't go down, so they hide decay | Plot rates, conversions, and cohort retention instead |
| Copying another company's north star | Metrics encode model mechanics you don't share | Derive the OMTM from your model x stage |
| Skipping cohorts | Blended averages mask whether the product improves | Track each signup cohort separately over time |
| Optimizing virality before stickiness | Growth multiplies churn -- the leaky bucket | Pass the retention gate, then build invite loops |
| Measuring what's easy, not what's risky | Decisions still get made on gut | Instrument the riskiest assumption first |
| No line in the sand | Every result gets rationalized; experiments can't fail | Pre-commit target, date, and miss response |
| Confusing correlation with causation | You pump a metric that doesn't drive the outcome | Run a controlled experiment before committing budget |

## Applications by Section

| Context | Application | Example |
|---------|-------------|---------|
| Dashboard audit | Rewrite totals as ratios | Total signups -> % of visitors activating within 7 days |
| Quarterly planning | One OMTM per stage; experiments ladder up to it | Stickiness stage -> all bets target week-4 retention |
| New product instrumentation | Name the model, install its metric tree | Subscription box -> primary model SaaS; churn tracked before AOV |
| Growth-spend decision | Check the stickiness gate first | D30 retention at 4% -> fix onboarding before buying ads |
| Target setting | Baseline -> line in the sand -> pre-commitment | "Churn under 4% by Q3 or we rebuild onboarding" |

## Quick Diagnostic

Have you named your business model and current stage? Is there exactly one OMTM, displayed everywhere? Does it have a paired counter-metric that prevents gaming? Is there a line in the sand (target, date, miss response)? Is data cohorted and segmented, not blended? Would every metric on the dashboard change a decision if it moved?