Product Metrics & Analytics

Product Metrics & Analytics

"If you can't measure it, you can't improve it." But the more common failure in product management is measuring the wrong things, or measuring things correctly but drawing the wrong conclusions.

Metrics discipline means defining what success looks like before a feature ships, selecting metrics that reflect actual user value rather than activity, and distinguishing between correlation and causation in the data.

The AARRR Framework

Dave McClure's Pirate Metrics maps the user lifecycle to five measurements:

Acquisition: How do users find you? Volume and cost by channel. Not all acquisition is equal — users from paid ads convert differently from users from word of mouth.

Activation: Do users have a great first experience? The activation rate is the percentage of new users who reach the "aha moment" — the first experience of core value. This is the most actionable metric in the early lifecycle.

Retention: Do users come back? Day 7 retention, Day 30 retention, cohort retention curves. The shape and height of the retention curve is the strongest signal of PMF.

Referral: Do users tell others? NPS as a leading indicator; new user attribution to word-of-mouth channels as a lagging one. A product with high referral rates grows without proportional increases in marketing spend.

Revenue: Do users pay? LTV, ARPU, MRR, churn. Revenue metrics confirm that users value the product enough to pay — but they are lagging indicators of product health.

North Star Metric

The North Star Metric captures the single best measurement of the value the product delivers to users. It sits between leading indicators and revenue:

• Spotify: daily listening minutes (time spent with the product doing what it is for)
• Airbnb: nights booked (the transaction that delivers value to both supply and demand sides)
• Slack: messages sent between colleagues (the action that makes the product's value tangible)
• Medium: total reading time (quality engagement, not just visits)

North Star properties:

• Directly reflects value delivery to users
• Leads revenue (high North Star value predicts future revenue)
• Is influenceable by product decisions
• Is a single number the entire team can understand and rally around

Churn and Retention

For subscription businesses, churn is the governing metric. A 5% monthly churn sounds small; it means losing 46% of customers per year. A business that acquires 1,000 new customers/month but churns 50/month (5%) will never grow beyond 1,000 customers regardless of how good its marketing is.

Customer churn: percentage of customers who cancel in a period. Problematic metric: can be stable while revenue churns (if small customers cancel while large ones stay).

Revenue churn: percentage of MRR lost to cancellations and downgrades. Net revenue retention includes expansion (upsells) — net NRR above 100% means the revenue from an existing customer cohort is growing even as some customers churn.

Cohort retention: tracking groups of users who started in the same period. Improving cohort retention curves (newer cohorts retain better than older ones) is the most direct evidence that product improvements are working.

LTV:CAC — The Unit Economics Test

LTV (Lifetime Value): average revenue per customer × gross margin / churn rate. A customer paying $50/month with 75% gross margin and 5% monthly churn has LTV = $50 × 0.75 / 0.05 = $750.

CAC (Customer Acquisition Cost): total marketing and sales spend / new customers acquired in the period.

LTV:CAC ratio: above 3:1 is generally considered healthy for SaaS. Below 1:1 means you are spending more to acquire customers than you'll ever earn from them.

The LTV:CAC ratio changes as the product scales — early-stage companies often have poor unit economics that improve as organic acquisition increases. But a product that cannot project a path to 3:1+ has a fundamental pricing or retention problem.

Metrics Traps

Vanity metrics: page views, registered users, social followers — can grow while the business declines. Always test: "What decision would this metric change?"

Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure." If engineers are measured on velocity (story points per sprint), they inflate estimates. If support agents are measured on ticket close rate, they close tickets without resolving issues.

Feature success metrics defined after the fact: teams tend to define success metrics after a feature ships, selecting metrics that make the feature look good. Define success metrics before building.

Attribution errors: the feature that shipped at the same time as a marketing campaign will appear to drive the metric movement the campaign drove. Correlation in time series data is not causation.