How to Define and Implement a North Star Metric Using PM Skills

PM Skills provides a dedicated North Star Metric skill that guides product teams through selecting, validating, and operationalizing a single customer-centric KPI using the /north-star command.

PM Skills is an open-source product management toolkit hosted at phuryn/pm-skills. The repository ships a specialized North Star Metric skill that automates the entire workflow of picking, validating, and operationalizing your primary KPI through a structured LLM-powered prompt.

Architecture of the North Star Metric System

The implementation consists of three distinct layers that work together to generate actionable metric frameworks.

Skill Definition Layer

The core logic resides in pm-marketing-growth/skills/north-star-metric/SKILL.md. This file contains the prompt template that instructs the model to classify your business game, suggest North Star candidates, and identify supporting input metrics.

Command Wrapper Layer

The pm-marketing-growth/commands/north-star.md file defines the /north-star slash-command. This wrapper collects your product description, forwards it to the skill, and formats the LLM response as a structured markdown report.

Execution Engine Layer

The generic pm-toolkit runtime parses commands, injects $ARGUMENTS into skill prompts, calls the LLM, and streams the rendered markdown back to the user. Reference the pm-toolkit/README.md for implementation details.

Step-by-Step Workflow

When you run the North Star Metric command, PM Skills executes a five-step pipeline grounded in product management best practices.

Step 1: Invoke the Command

Trigger the skill using the slash-command syntax defined in north-star.md (lines 10-30):

/north-star B2B SaaS for team collaboration

The command accepts any product description as its argument.

Step 2: Inject Context

The execution engine replaces the $ARGUMENTS placeholder in SKILL.md (line 37) with your supplied description. This creates a complete prompt that provides the LLM with full business context.

Step 3: Classify and Validate

The skill runs three sequential operations:

• Business Game Classification: Categorizes your product into one of three games—Attention, Transaction, or Productivity—using the definitions at lines 27-31 of SKILL.md
• Candidate Generation: Proposes 2-3 North Star metric candidates
• Validation: Tests each candidate against seven criteria (lines 42-51) including "Easy to Understand," "Customer-Centric," and "Sustainable Value"

Step 4: Extract Input Metrics

For the selected North Star, the skill identifies 3-5 input metrics that directly drive the outcome (lines 53-57). These represent the ownable levers your team can pull to influence the North Star.

Step 5: Render the Report

The command template (lines 59-98) assembles a markdown document containing:

• Validation tables for each criterion
• Input metrics with definitions
• A metrics constellation diagram
• Next-step recommendations

Practical Implementation Examples

Basic Command Usage

Run the command from any PM Skills interface:

/north-star Consumer fitness app monetized through subscriptions

The output follows the template defined in north-star.md:

## North Star Framework: Consumer fitness app

**Business Game**: Productivity

### North Star Metric

**Metric**: Weekly Active Workouts Completed
**Definition**: Number of distinct users who log a completed workout at least once per week

| Criterion            | Pass? | Notes                                         |
|----------------------|-------|-----------------------------------------------|
| Easy to Understand   | Y     | Clear action ("completed workout")            |
| Customer-Centric     | Y     | Reflects value delivered to user              |
| Sustainable Value    | Y     | Correlates with long-term retention         |

Programmatic Invocation

Embed the skill in a custom Node.js bot using the toolkit API:

import { runCommand } from '@pm-toolkit/cli';

async function defineNorthStar(productDesc) {
  const result = await runCommand('/north-star', productDesc);
  return result.markdown;   // Structured report ready for display
}

defineNorthStar('Marketplace for freelance designers');

The runCommand helper internally loads north-star.md, substitutes $ARGUMENTS, and returns the structured markdown payload.

Integrating with OKRs

Copy the generated tables directly into your OKR tracker:

# Q3 OKR – Growth

**Objective:** Grow product-value adoption

- **Key Result 1:** Increase Weekly Active Workouts (North Star) from 5,000 to 8,000
- **Key Result 2:** Boost "Workout Completion Rate" (input metric) by 15%
- **Key Result 3:** Reduce "Time-to-First-Workout" (counter-metric) to < 24 hours

Summary

• Customer-centric focus: The SKILL.md prompt enforces that your North Star captures value delivered to the customer, not vanity metrics or raw revenue.
• Actionable levers: The requirement for 3-5 input metrics provides a clear map of ownable metrics that drive the North Star.
• Repeatable process: The /north-star command can be run quarterly or whenever your business model shifts, ensuring the metric stays aligned with product strategy.
• Three-layer architecture: Implementation spans pm-marketing-growth/skills/north-star-metric/SKILL.md, pm-marketing-growth/commands/north-star.md, and the pm-toolkit execution engine.

Frequently Asked Questions

What are the three business games used for classification?

The skill classifies products into Attention (content/engagement platforms), Transaction (marketplaces/e-commerce), or Productivity (tools that help users create output) as defined in SKILL.md lines 27-31. This classification ensures the proposed North Star matches your fundamental value proposition.

How does PM Skills validate North Star candidates?

Each candidate is evaluated against seven criteria explicitly defined in the skill prompt: Easy to Understand, Customer-Centric, captures Sustainable Value, Actionable, Measurable, not a Vanity Metric, and aligned with the Business Game. The validation table in the output shows pass/fail status for each criterion with explanatory notes.

Can I customize the number of input metrics generated?

The skill template in SKILL.md (lines 53-57) is optimized to return 3-5 input metrics. While you can modify the source file to request more or fewer metrics, the current implementation targets this range to maintain focus on the highest-impact levers without overwhelming the team with too many variables.

Where does the LLM prompt live in the repository?

The core prompt logic resides in pm-marketing-growth/skills/north-star-metric/SKILL.md. The command wrapper that handles user input and formatting is located at pm-marketing-growth/commands/north-star.md. The execution engine that processes these files is documented in pm-toolkit/README.md.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}