Principles of Product Management: Metric Trees and North Star Metric (Input vs Output Metrics, Lagging vs Leading)

Metric Trees and North Star Metric (Input vs Output Metrics, Lagging vs Leading)

Metric Trees & North Star Metric (Input vs Output, Lagging vs Leading)

What This Is

A metric tree is a hierarchical breakdown of how your product’s high-level goals (e.g., revenue, retention) cascade into smaller, actionable metrics. Your North Star Metric (NSM) is the single, most critical metric that best captures the core value your product delivers to users. Together, they help teams focus on what actually drives long-term success—not just vanity metrics. Example: For Duolingo, the NSM is "Weekly Active Learners" (output), but the metric tree includes input metrics like "lessons completed per session" (leading) and "7-day retention" (lagging).

Key Terms & Frameworks

• North Star Metric (NSM): The one metric that best reflects the value your product delivers. It should align with business goals, be actionable, and predict long-term success. Example: Airbnb’s NSM is "Nights Booked" (not "DAU" or "revenue").

• Metric Tree: A visual hierarchy that breaks down your NSM into input metrics (levers you control) and output metrics (results you measure). Structure: NSM-Output Metrics-Input Metrics-Experiments/Features

• Input vs Output Metrics:

• Input (Leading): Metrics you can directly influence (e.g., "onboarding completion rate").

• Output (Lagging): Metrics that reflect outcomes (e.g., "revenue per user"). Input metrics predict output metrics.

• Lagging vs Leading Indicators:

• Lagging: Historical results (e.g., "churn rate last month").

• Leading: Predictive signals (e.g., "users who complete onboarding are 2x less likely to churn").

• HEART Framework (Google): A way to categorize metrics for user experience: Happiness (NPS), Engagement (sessions/week), Adoption (new users), Retention (7-day retention), Task Success (completion rate).

• AARRR (Pirate Metrics): Funnel stages for growth: Acquisition-Activation-Retention-Revenue-Referral.

• ICE Score (Prioritization): Impact × Confidence × Ease – used to rank experiments/features.

• Contribution Analysis: Quantifying how much each input metric contributes to the NSM (e.g., "A 1% increase in onboarding completion-0.5% increase in retention").

• Vanity Metrics: Metrics that look good but don’t correlate with business outcomes (e.g., "total app downloads" vs "active users").

• Counter Metrics: Metrics that ensure you’re not optimizing for the NSM at the expense of other critical goals (e.g., "increasing engagement" but "decreasing NPS").

Step-by-Step Process Flow

1. Define Your NSM
2. Ask: "What’s the single metric that best captures the value we deliver?"
3. Example: For a fitness app, it might be "Weekly Active Users Who Log a Workout" (not just "DAU").

4. Test: Does improving this metric correlate with long-term success (e.g., retention, revenue)?

5. Build the Metric Tree

6. Start with the NSM at the top.
7. Break it into output metrics (e.g., "retention rate," "revenue per user").
8. Break those into input metrics (e.g., "onboarding completion rate," "session frequency").

9. Example Tree for a SaaS Product: NSM: Monthly Active Teams (MAT) -Output: Retention Rate, Revenue per Team -Input: Onboarding Completion, Feature Adoption, Support Tickets

10. Identify Leading vs Lagging Indicators

11. Leading (Input): "Users who complete onboarding in <2 mins" (predicts retention).
12. Lagging (Output): "30-day retention rate" (measures past success).

13. Action: Focus experiments on leading indicators—they’re your levers.

14. Run Contribution Analysis

15. Use regression analysis or A/B tests to quantify how input metrics impact the NSM.

16. Example: "A 10% increase in onboarding completion-3% increase in 7-day retention."

17. Prioritize Experiments with ICE

18. Score potential experiments (e.g., "improve onboarding") using Impact × Confidence × Ease.

19. Example: "Redesign onboarding" might score 8 × 7 × 5 = 280.

20. Monitor Counter Metrics

21. Track metrics that could be harmed by optimizing for the NSM.
22. Example: If your NSM is "time spent in app," counter metrics might be "NPS" or "support tickets."

Common Mistakes

• Mistake: Choosing a vanity metric as the NSM (e.g., "total users" instead of "active users"). Correction: The NSM should reflect real user value, not just growth. Example: Facebook’s early NSM was "7 friends in 10 days" (not "DAU").

• Mistake: Ignoring lagging indicators (e.g., only tracking "signups" but not "retention"). Correction: Lagging metrics validate whether your leading indicators are working. Example: If "onboarding completion" (leading) improves but "30-day retention" (lagging) doesn’t, your onboarding isn’t driving real value.

• Mistake: Building a metric tree without input metrics (e.g., only tracking "revenue" but not "feature adoption"). Correction: Input metrics are your levers—without them, you can’t take action. Example: If "revenue" is down, you need to know if it’s due to "lower conversion" or "fewer users."

• Mistake: Optimizing for the NSM at the expense of everything else (e.g., increasing "engagement" but hurting "NPS"). Correction: Always track counter metrics to avoid unintended consequences.

• Mistake: Assuming correlation = causation (e.g., "users who watch tutorials have higher retention"-"force everyone to watch tutorials"). Correction: Run experiments to prove causation. Example: A/B test "mandatory tutorial" vs "optional tutorial" to see if it actually improves retention.

PM Interview / Practical Insights

1. Tricky Distinction: Input vs Output Metrics
2. Interviewer Trap: "Why not just track revenue as your NSM?"

3. Answer: Revenue is an output metric—it’s a result, not a lever. The NSM should be something you can influence directly (e.g., "transactions per user" for a marketplace).

4. Leading vs Lagging Indicators

5. Interviewer Trap: "How would you measure the success of a new onboarding flow?"

6. Answer: Track leading indicators like "onboarding completion rate" (predicts retention) and lagging indicators like "7-day retention" (validates success).

7. Counter Metrics

8. Interviewer Trap: "Your team wants to increase engagement by adding push notifications, but NPS is dropping. What do you do?"

9. Answer: Pause and investigate. Engagement is good, but if NPS (a counter metric) is dropping, the notifications might be annoying users. Run a survey to understand why.

10. Metric Tree Depth

11. Interviewer Trap: "How granular should a metric tree be?"
12. Answer: Deep enough to be actionable, but not so deep that it’s overwhelming. Rule of thumb: 3–4 levels max (NSM-Output-Input-Experiments).

Quick Check Questions

1. Scenario: Your team wants to launch a feature that increases "time spent in app" (your NSM) but decreases "NPS." How do you decide? Answer: Don’t launch. The NSM should align with user value—if NPS drops, the feature isn’t delivering real value. Why: Counter metrics protect against short-term optimization.

2. Scenario: You’re the PM for a food delivery app. Your NSM is "Weekly Active Users." What are 2 leading and 2 lagging indicators you’d track? Answer:

3. Leading: "First-time order completion rate," "Average delivery time."

4. Lagging: "7-day retention rate," "Revenue per user." Why: Leading indicators predict future success; lagging indicators validate it.

5. Scenario: Your CEO says, "Let’s make ‘total app downloads’ our NSM." How do you respond? Answer: Push back. "Total downloads" is a vanity metric—it doesn’t reflect active usage or value. Suggest "Weekly Active Users" or "Retention Rate" instead. Why: The NSM should measure real engagement, not just growth.

Last-Minute Cram Sheet

1. NSM = The one metric that best captures user value (e.g., "Nights Booked" for Airbnb).
2. Metric Tree = NSM-Output Metrics-Input Metrics-Experiments.
3. Input (Leading) Metrics = Levers you control (e.g., "onboarding completion").
4. Output (Lagging) Metrics = Results you measure (e.g., "revenue").
5. HEART Framework = Happiness, Engagement, Adoption, Retention, Task Success.
6. AARRR = Acquisition, Activation, Retention, Revenue, Referral.
7. ICE Score = Impact × Confidence × Ease (prioritize experiments).
8. Counter Metrics = Track these to avoid optimizing for the NSM at all costs.
9. Vanity Metrics = "Total users," "page views"—don’t use these as NSMs.
10. Correlation-Causation – Always run experiments to prove impact.