PMF OS
Your AI Product-Market Fit Coach.
PMF OS helps you find product-market fit after launch. It diagnoses why users leave, finds patterns that predict retention, and guides you to PMF.
Price: $99 (one-time)
Quick Start
npx @aiorg/cli init pmf-os ~/my-project
cd ~/my-project
claude
> /setup
Overview
PMF OS doesn't just tell you "156 users are inactive." It helps you understand WHY they left and what to do about it.
| Traditional Analytics | PMF OS |
|---|---|
| "156 users inactive" | "Users leave because they never experience value" |
| Dashboard of metrics | Actionable diagnosis with next steps |
| You interpret data | Claude interprets and recommends |
| Same for everyone | Adapts to your stage and data |
When to Use PMF OS
Use PMF OS if:
- Users sign up but don't come back
- You don't know why users leave
- You're not sure if you have product-market fit
- You're considering a pivot
Don't use PMF OS if:
- You haven't launched yet → Use Idea OS
- Users activate but churn later → Use Success OS
- You need more traffic → Use Marketing OS
What's Included
Commands
Setup & Data
| Command | Description |
|---|---|
/setup | Connect to project, import shared context |
/import | Import user data (CSV/JSONL) |
/sync | Refresh data, check for updates |
Diagnosis
| Command | Description |
|---|---|
/diagnose | Root cause analysis (why users leave) |
/activation | Activation rate deep dive |
/funnel | Where users drop off |
/cohorts | Retention by signup week |
Pattern Discovery
| Command | Description |
|---|---|
/patterns | Retained vs churned comparison |
/segments | Who are your best users? |
/magic | Find the "magic number" |
User Research
| Command | Description |
|---|---|
/interview | Guide to user interviews |
/survey | Create surveys (Sean Ellis, etc.) |
/feedback | Analyze feedback patterns |
PMF Measurement
| Command | Description |
|---|---|
/pmf | Product-market fit assessment |
/score | Track PMF score over time |
/ready | Am I ready to scale? |
Experiments
| Command | Description |
|---|---|
/experiment | Define and track experiments |
/hypothesis | Generate hypotheses from data |
/results | Analyze experiment results |
Planning
| Command | Description |
|---|---|
/sprint | 30-day PMF sprint |
/weekly | Weekly review |
/next | What to focus on now |
Core Concepts
Activation vs Retention
Activation = User experiences core value (first time) Retention = User keeps coming back (ongoing)
If activation is broken, retention doesn't matter. You can't retain users who never activated.
The Magic Number
Every successful product has a "magic number" — an action that predicts retention:
- Facebook: 7 friends in 10 days
- Slack: 2,000 messages
- Dropbox: 1 file in 1 folder
PMF OS helps you find yours through /patterns analysis.
Sean Ellis Test
The gold standard for PMF:
"How would you feel if you could no longer use [product]?"
- Very disappointed
- Somewhat disappointed
- Not disappointed
40%+ "Very disappointed" = PMF achieved
Run this survey via /survey.
Session Start Behavior
When you open PMF OS, it doesn't wait for commands. It immediately:
- Checks your project connection
- Analyzes your state
- Shows a PMF dashboard:
PMF DASHBOARD
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PMF Score: Not measured yet
Activation Rate: 23% (needs work)
Week 1 Retention: 45% (approaching healthy)
My Diagnosis:
├── Your #1 problem is ACTIVATION
│ 73% of users never complete [core action]
│
├── Users who [action] in first 24h retain 3x better
│ This is your magic number candidate
│
└── Hypothesis: Onboarding doesn't guide to value
Recommended Actions:
1. [URGENT] Run /diagnose for root cause analysis
2. [HIGH] Interview 5 churned users (/interview)
3. [MEDIUM] Measure PMF score (/pmf)
Stage-Aware Advice
< 100 Users
With [X] users, every conversation matters.
I recommend:
├── Interview at least 10 churned users
├── Manual analysis of every signup
├── Focus on UNDERSTANDING, not metrics
└── Don't worry about statistical significance
100-1000 Users
With [X] users, patterns emerge.
I recommend:
├── Segment analysis (who retains vs churns)
├── Magic number discovery
├── First experiments
└── Sean Ellis survey
> 1000 Users
With [X] users, you can be rigorous.
I recommend:
├── A/B tests with statistical significance
├── Cohort analysis
├── Automated surveys
└── Formal experiment tracking
Data Structure
PMF OS stores data in .pmf-os/ in your project:
.pmf-os/
├── contacts.db # User data (SQLite)
├── experiments/ # Experiment definitions and results
├── interviews/ # Interview notes and insights
├── surveys/ # Survey templates and responses
├── reports/ # Generated reports
└── config.json # PMF OS configuration
Shared Context
PMF OS reads and writes to the shared context layer (~/.aiorg/projects/):
What it reads:
- Idea OS validation score
- Target customer from Idea OS
- Business stage
What it writes:
- PMF score (0-100)
- PMF status (searching | approaching | achieved)
- Sean Ellis score
- Activation rate
Related Kits
Before:
- Idea OS — Validate before building
- SaaS Dev Team — Build your product
After:
- Marketing OS — Scale user acquisition (after PMF)
- Success OS — Retain activated users
FAQ
Q: What's the difference between PMF OS and Success OS?
PMF OS: Users leave before activating (product problem) Success OS: Users leave after activating (retention problem)
Q: How do I know if I have PMF?
Run /pmf for a full assessment. Key indicators:
- 40%+ "very disappointed" on Sean Ellis survey
- Organic growth (users referring others)
- High activation rate (>50%)
Q: What if I don't have user data?
PMF OS can work with qualitative data. Run /interview to structure user conversations and analyze patterns manually.
Q: Can PMF OS help me pivot?
Yes. If diagnosis suggests a pivot, PMF OS helps you form hypotheses and test new directions.