Product Metrics Explained: North Star Metric, KPIs, DAU, MAU and Retention

A product manager who cannot decompose their product's behavioral telemetry mathematically is flying blind. Yet many product dashboards are cluttered with vanity metrics—cumulative app downloads, total registered accounts since 2022, or raw pageviews—that go up and to the right every week even as active user retention silently collapses. Conversely, relying solely on top-line financial accounting metrics like Annual Recurring Revenue (ARR) gives product squads a lagging indicator that reacts months after a bad product release has already alienated customers.

As codified in Amplitude's North Star Framework and Google Analytics 4 (GA4) Engagement Documentation, world-class product analytics connects customer value delivery to sustainable business revenue through a structured Metric Tree. At the apex sits the North Star Metric (NSM)—a single leading indicator that captures the exact moment users realize core value from the product—supported underneath by controllable squad-level Input Metrics across Acquisition, Activation, Engagement ($\text{DAU}/\text{MAU}$), Retention, and Monetization. This guide defines the exact mathematical formulas for every foundational product metric, contrasts the three types of cohort retention (N-Day, Unbounded, and Bracket), explains how to build a decomposed Metric Tree, and walks through a complete quantitative growth-accounting and retention audit.

1. The North Star Metric (NSM) and Its Four Input Levers

Per Amplitude's North Star Playbook (and What Makes a Good vs. Bad North Star Metric), a valid North Star Metric must satisfy three strict criteria: (1) it must measure the moment the customer receives core value (not merely opening the app or paying an invoice), (2) it must lie within the product and engineering team's direct sphere of influence, and (3) it must act as a leading indicator of long-term revenue and retention.

Why is "Monthly Revenue (MRR)" a poor North Star Metric for a product squad? Because a squad can temporarily boost short-term MRR by hiding the "Cancel Subscription" button or raising prices 30% overnight—actions that destroy customer trust and cause churn two quarters later. Instead, choose a North Star aligned with your product's Core Engagement Game:

• Attention Game (Media / Streaming / Social): Total hours of engaged streaming per week or Daily active creators publishing content.
• Transaction Game (E-Commerce / Fintech / Ride-Sharing): Weekly completed cross-border settlements or Monthly rides completed without cancellation.
• Productivity Game (B2B SaaS / Developer Tools / Analytics): Weekly active teams completing $\ge 5$ automated workflows or Weekly queries executed returning verified insights (Amplitude Defining Your North Star).

To make the North Star actionable for individual squads, decompose it mathematically into Four Input Levers:

1. Breadth: How many new and returning users/workspaces enter the active loop?
2. Depth: How deeply does each user engage per session (e.g., features adopted or team seats invited)?
3. Frequency: How often does each user return to complete the core action per week/month?
4. Efficiency / Speed: What percentage of initiated attempts succeed rapidly without error?

2. Exact Mathematical Formulas for Core Product Metrics

Every Product Manager must know these exact definitions and formulas cold—both for building SQL/BI dashboards (Topical Knowledge Hubs — Data Science & BI) and for answering analytical interview cases (Article 21, Article 24):

1. Active Users ($\text{DAU}$, $\text{WAU}$, $\text{MAU}$) & Stickiness Ratio:
   Crucially, "Active" must never be defined as "logged in or loaded the homepage." Define Active User as a unique user who completes at least one meaningful value event within the window (1 day, 7 days, or 30 days): $$\text{Stickiness Ratio} = \frac{\text{DAU (Daily Active Users)}}{\text{MAU (Monthly Active Users)}} \times 100%$$ (Interpretation: A $\text{DAU}/\text{MAU}$ ratio of $20%$ means the average monthly active user engages on $0.20 \times 30 = \mathbf{6\text{ days per month}}$. Consumer social/messaging apps target $50%\text{–}70%+$; B2B daily workflow SaaS targets $25%\text{–}45%$ when measured on working weekdays).

2. The Three Mathematical Variants of Cohort Retention (Amplitude Metrics Guide):
   Given a starting cohort of $N_0$ users who performed their first key action on Day 0:

   • Classic N-Day (Exact-Day) Retention: The percentage of $N_0$ who return and perform a value action on exact Day $k$ (e.g., Day 7 = 7 days after signup):
     $$\text{N-Day Retention}_k = \frac{\text{Users from } N_0 \text{ active on exact Day } k}{N_0}$$ (Best for high-frequency daily habit products like mobile games or consumer messaging).
   • Bracket (Bounded Window) Retention: The percentage of $N_0$ who return at least once within a multi-day interval bucket (e.g., Week 2 = Days 8–14; Month 2 = Days 31–60):
     $$\text{Bracket Retention}_{[t_1, t_2]} = \frac{\text{Users from } N_0 \text{ active at least once in } [t_1, t_2]}{N_0}$$ (Best for weekly or monthly B2B SaaS workflows and fintech platforms).
   • Unbounded (Rolling / Return-On-or-After) Retention: The percentage of $N_0$ who return on Day $k$ or any day after Day $k$:
     $$\text{Unbounded Retention}_k = \frac{\text{Users from } N_0 \text{ active on any day } t \ge k}{N_0}$$ (Mathematically equal to $1 - \text{True Permanent Churn}_k$; best for low-frequency products).

3. Growth Accounting & The SaaS Quick Ratio:
   Any change in Monthly Active Users ($\Delta \text{MAU}$) or Monthly Recurring Revenue ($\Delta \text{MRR}$) obeys the conservation identity: $$\Delta \text{MAU}_t = \underbrace{\text{New Users}_t + \text{Resurrected Users}t}{\text{Gross Inflows}} - \underbrace{\text{Churned Users}t}{\text{Gross Outflows}}$$ $$\text{User Quick Ratio} = \frac{\text{New Users}_t + \text{Resurrected Users}_t}{\text{Churned Users}_t}$$ (A Quick Ratio $= 1.0$ means the product is treading water; healthy growth requires a Quick Ratio $> 1.5\text{–}2.0$ for consumer users and $> 3.0\text{–}4.0$ for B2B SaaS MRR).

4. Net Revenue Retention (NRR) vs. Gross Revenue Retention (GRR): $$\text{NRR} = \frac{\text{Starting ARR} + \text{Expansion ARR} - \text{Contraction ARR} - \text{Churned ARR}}{\text{Starting ARR}} \times 100%$$ $$\text{GRR} = \frac{\text{Starting ARR} - \text{Contraction ARR} - \text{Churned ARR}}{\text{Starting ARR}} \times 100%\quad (\text{Bounded }\le 100%)$$

Frameworks and Comparison Tables

Table 17.1: Vanity Metrics vs. Actionable Product Metrics Across the Pirate (AARRR) Funnel

Table 17.2: North Star Metric & Input Tree Examples Across Four Product Archetypes

Visualizing the Framework

The following architecture diagram illustrates the workflow and decision criteria:

flowchart TD
    NSM["North Star Metric: Weekly Automated Invoices Reconciled = 180,000/wk"] --> L1[Breadth: Active Transacting Workspaces = 6,000]
    NSM --> L2["Frequency x Depth: Avg Invoices Uploaded per Workspace/wk = 33.3"]
    NSM --> L3["Efficiency: Automated Match Success Rate = 90.0%"]
    L1 --> S1[Onboarding Squad Lever: New Signups x Day-7 Activation Rate]
    L1 --> S2[Retention Squad Lever: Prior Cohort Survival Rate]
    L2 --> S3[Core Workflow Squad Lever: Bulk Bank-Feed Auto-Sync Adoption]
    L3 --> S4["AI Extraction Squad Lever: OCR & Schema Validation Accuracy"]

Practical Worked Example

Scenario (Hypothetical Prosumer AI Productivity App — NoteGenius):
The CEO of NoteGenius celebrates at the monthly all-hands because Monthly Active Users ($\text{MAU}$) grew from $50,000$ in January to $58,000$ in April ($+16%$) follow-on growth. However, as the newly hired Product Manager, you decompose the underlying Growth Accounting, Stickiness, and Cohort Retention tables to verify whether the product is genuinely healthy.

Step 1 — Decomposing April's Growth Accounting & User Quick Ratio:
Querying the April event database reveals how $\text{MAU}$ moved from March ($\text{MAU}{\text{Mar}} = 55,000$) to April ($\text{MAU}{\text{Apr}} = 58,000$):

• New Activated Users in April (driven by a $120k paid influencer blitz): $+18,000$
• Resurrected Users in April: $+2,000$
• Churned Users in April (Active in March, 0 sessions in April): $-17,000$ ($30.9%$ monthly logo churn!)
• Net $\Delta \text{MAU}$: $+18,000 + 2,000 - 17,000 = +3,000$ ($\text{MAU}_{\text{Apr}} = 58,000$).
• Calculating the User Quick Ratio: $$\text{Quick Ratio}_{\text{Apr}} = \frac{\text{New} + \text{Resurrected}}{\text{Churned}} = \frac{18,000 + 2,000}{17,000} = \frac{20,000}{17,000} = \mathbf{1.18\times}$$ Finding 1: A Quick Ratio of $1.18\times$ with $30.9%$ monthly churn proves NoteGenius is a leaky bucket: the moment marketing pauses the $120k monthly ad spend, $\text{MAU}$ will collapse by ~30% per month!

Step 2 — Comparing Classic N-Day vs. Bracket vs. Unbounded Retention on the January Cohort ($N_0 = 10,000$ Signups):

• Users active on exact Day 30: $420\text{ users} \rightarrow \text{Exact Day-30 Retention} = \mathbf{4.2%}$.
• Users active at least once during Month 2 (Days 31–60): $1,450\text{ users} \rightarrow \text{Month-2 Bracket Retention} = \mathbf{14.5%}$.
• Users active at least once during Month 3 (Days 61–90): $810\text{ users} \rightarrow \text{Month-3 Bracket Retention} = \mathbf{8.1%}$ (Curve has not flattened; decaying by ~44% per month).
• Users active on Day 30 or any day after Day 30 (Unbounded Day-30): $1,720\text{ users} \rightarrow \text{Unbounded Retention}_{30} = \mathbf{17.2%}$ (meaning $82.8%$ of the January cohort is permanently dead after just 30 days).

Step 3 — Multiplicative North Star Input Leverage Modeling:
Currently, the North Star Metric (Weekly AI Summaries Saved & Shared) equals: $$\text{NSM}_0 = \text{WAU (14,500)} \times \text{Attempts/WAU (2.5)} \times \text{Completion Success Rate (64%)} = \mathbf{23,200\text{ summaries/week}}$$ Instead of spending another $120k on top-of-funnel ads, you reallocate Q3 product capacity to two retention & efficiency levers:

1. Fixing LLM timeout errors lifts Completion Success Rate from $64%$ to $85%$ ($+32.8%$ relative lift).
2. Adding an automated Calendar Meeting Bot lifts Weekly Active Users ($\text{WAU}$) from $14,500$ to $19,000$ ($+31.0%$ lift) and Attempts/WAU from $2.5$ to $3.2$ ($+28.0%$ lift). $$\text{New NSM}_1 = 19,000 \times 3.2 \times 0.85 = \mathbf{51,680\text{ summaries/week}}\ (+122.8%\text{ compound lift without increasing ad spend})$$

Common Mistakes and Limitations

1. Defining "Active User" as Any App Open or Login: If an angry customer logs insolely to download their past invoices and cancel their account, a naive tracker counts them as an "Active User" for that month! Always anchor $\text{DAU}/\text{MAU}$ and Retention in a verified core value event (e.g., query_executed, transaction_settled, document_exported).
2. Averaging Retention Across All Cohorts into a Single Blended Number: A single blended retention percentage hides whether your product improvements over the last three months actually made the product better for new users. Always plot triangular cohort retention tables by signup week/month so you can compare whether the October cohort retains at a higher Day-7 rate than the July cohort.
3. Goodhart's Law ("When a Measure Becomes a Target, It Ceases to Be a Good Measure"): Every single metric can be gamed if tracked without a guardrail counter-metric (Article 10). Always pair engagement/conversion metrics with quality, latency, and churn guardrails.

Practical Implementation Checklist

• [ ] North Star Criteria Check: Does our North Star Metric measure customer value realization rather than raw revenue or superficial pageviews?
• [ ] Mathematical Tree Decomposition: Have we decomposed our North Star Metric into mutually exclusive, collectively exhaustive (MECE) input levers across Breadth, Depth, Frequency, and Efficiency?
• [ ] Value-Event Active Definition: Is our definition of $\text{DAU}/\text{WAU}/\text{MAU}$ tied strictly to completing a core value action rather than loading a page or receiving a push notification?
• [ ] Right Retention Formula Selected: Are we using Bracket Retention for weekly/monthly workflows, N-Day Retention for daily habits, and Unbounded Retention for churn audits?
• [ ] Growth Accounting Separation: Does our executive dashboard separate New, Retained, Resurrected, and Churned users via the Quick Ratio so paid acquisition never masks churn?

Frequently Asked Questions

1. Can a company have more than one North Star Metric?

A single product line or business unit should have one overarching North Star Metric so that Product, Design, Engineering, and Growth align around a single shared equation (Amplitude North Star Hub). Only a multi-product conglomerate operating completely independent business models (for example, Uber Rides vs. Uber Freight, or Atlassian Jira vs. Trello) should maintain separate North Star Metrics per business unit—and even then, each individual product squad owns a specific Input Metric on that unit's tree rather than inventing an isolated North Star.

2. How did Google Analytics 4 (GA4) change the definition of "Bounce Rate" and "Engagement Rate"?

In legacy Universal Analytics, a user who clicked a blog article, read every word for 4 minutes, found their exact answer, and closed the tab without clicking a second page was counted as a $100%$ "Bounce"! Per Google Analytics 4 Official Documentation (and GA4 Engagement Metrics Guide), GA4 replaced that flawed definition with Engaged Sessions—defined as any session that lasts longer than 10 seconds (configurable up to 60 seconds), has a key conversion event, or has $\ge 2$ page/screen views. In GA4, $\text{Engagement Rate} = \frac{\text{Engaged Sessions}}{\text{Total Sessions}}$, and $\text{Bounce Rate}$ is simply the mathematical inverse ($100% - \text{Engagement Rate}$).

3. What is a good DAU/MAU Stickiness ratio benchmark?

Benchmarks depend entirely on the natural frequency of the customer's Job to Be Done:

• Daily Communication / Social / Messaging Apps (WhatsApp, Slack): $50%\text{ to }80%+$
• B2B Daily SaaS Workflows (Issue Trackers, CRM, IDE Copilots): $25%\text{ to }45%$ (Note: because B2B users work 5 days a week—roughly 21–22 workdays per 30-day month—the theoretical ceiling for a weekday-only B2B tool is $\approx 72%$, or compute Workday Stickiness excluding weekends).
• E-Commerce / On-Demand Delivery: $10%\text{ to }20%$
• Travel / Tax / Mortgage Platforms: $\text{DAU}/\text{MAU}$ is the wrong metric; track quarterly or annual repeat transaction retention instead.

4. Why can Net Revenue Retention (NRR) exceed 100%, and why do B2B investors obsess over it?

Gross Revenue Retention (GRR) cannot exceed $100%$ because it only measures how much of last year's starting ARR was retained after downgrades and churn. Net Revenue Retention (NRR) includes Expansion ARR (existing customers buying more seats, higher usage tiers, or cross-sell modules). If a B2B SaaS company has $\text{NRR} = 120%$, its existing customer base grows by $20%$ every year automatically—even if the sales team signs zero new logos!

5. What is the "Aha! Moment" (or Activation Metric), and how do you find it mathematically?

The Aha! Moment is the specific early behavioral threshold (e.g., "A new workspace that connects a data source and shares 1 chart with a colleague within 72 hours of signup") where a user experiences core value for the first time. To find it mathematically, plot a $2 \times 2$ contingency matrix comparing early behaviors (Days 1–7) against long-term retention (Month 3 Retained vs. Churned), maximizing both Precision/Lift (users who complete Action $X$ retain at $3\times$ the baseline rate) and Recall/Coverage (a large proportion of retained users completed Action $X$), then validate causality via an A/B experiment (Article 18).

6. How should I organize an Event Taxonomy so our analytics database doesn't become a mess?

Enforce a strict object_action snake_case naming convention documented in a shared tracking plan (e.g., invoice_created, invoice_pdf_downloaded, webhook_replay_clicked) paired with standardized Event Properties (account_tier, user_role, latency_ms, source_screen). Never create separate event names for property variations (use button_clicked with property location: header rather than header_blue_button_clicked_v2).

Next Steps and Recommended Reading

Product metrics are the sensory nervous system of a product organization. By anchoring your team in a value-based North Star Metric, decomposing it into controllable Breadth, Depth, Frequency, and Efficiency inputs, and tracking Cohort Retention and Quick Ratios instead of vanity aggregates, you gain the mathematical clarity to invest engineering effort where it compounds fastest.

Practical Next Step: Draw your product's North Star Metric Tree on a single page today (following the exact structure in Section F), plug in last month's actual numbers for each node, and identify which single input lever has the largest gap against benchmark.

Recommended Follow-Up Reading: Now that your metric tree and event taxonomy are defined, proceed to Article 18: A/B Testing for Product Managers: Hypotheses, Sample Size and Results (/blog/product/ab-testing-for-product-managers/ — Planned Destination) to learn how to prove causal metric lifts with statistical rigor.

About this guide

Aniket Thakur prepared this comprehensive Product Management guide based on continuous discovery frameworks, operational research, and practitioner benchmarks checked on 24 October 2026. All quantitative scenarios in Section G are explicitly labelled hypothetical worked examples illustrating decision math.