Pirate Metrics
Mapea tu producto en AARRR (Acquisition, Activation, Retention, Referral, Revenue) y diagnostica el embudo.
---
name: pirate-metrics
description: Use Dave McClure's AARRR funnel to diagnose where a product is leaking. For each stage (Acquisition, Activation, Retention, Referral, Revenue) define the canonical event, the current rate, the target rate, and the next experiment to run. Output a one-page funnel with the weakest stage flagged.
---Pirate Metrics (AARRR)
Coined by Dave McClure at 500 Startups. Five stages every product goes through. Most teams obsess over Acquisition and ignore Retention — which is where the real money lives.
When to use this skill
- New product, no analytics culture yet.
- Existing product where the team can't agree what's broken.
- Investor due diligence prep.
- Quarterly health check.
The five stages
Acquisition
The user lands and shows interest. Canonical event: landed_homepage or signed_up_for_newsletter. Healthy rate: depends on channel. SEO 1-3% click-to-signup; paid 0.5-2%; referral 8-15%.
Activation
The user has their first valuable experience. Canonical event varies — for Slack it's "sent 2000 messages in a team of 3+", for Dropbox "uploaded 1 file from 2 devices". Healthy rate: 30-50% of signups for B2B SaaS, 15-25% for consumer. This is the biggest leak in most products. Optimize here before paying for more traffic.
Retention
The user comes back. Canonical event: weekly or monthly active user, where active = the activation event repeated. Healthy rate: D1 60%+, D7 40%+, D30 25%+ for consumer SaaS. B2B varies but should be flat or rising at month 3+. Retention shape matters more than absolute number. Flat = product market fit. Decaying = leaks.
Referral
The user invites others. Canonical event: invited_a_friend and the friend signed up. Healthy rate: viral coefficient (k-factor) > 0.5 sustained. Most B2B products have k < 0.1; that's fine if Acquisition + Retention are strong.
Revenue
The user pays. Canonical event: first paid charge. Healthy rate: free → paid 2-5% for self-serve B2B; LTV/CAC > 3 sustained.
Process
- For each stage, define the canonical event in plain language.
- Pull current rate from analytics.
- Set a target rate based on benchmark or hypothesis.
- Identify the weakest stage — where the multiplicative gap is biggest.
- Propose 2-3 experiments to fix the weakest stage. Don't optimize the others until the leak is fixed.
Output format
| Stage | Event | Current | Target | Gap |
|---------------|--------------------------------------|---------|--------|--------|
| Acquisition | Visit landing | 12,400 | 12,400 | — |
| | → Signed up | 4.1% | 6.0% | -1.9pp |
| Activation | First report generated within 7d | 31% | 50% | -19pp | ← weakest
| Retention | Active 4+ weeks after activation | 64% | 70% | -6pp |
| Referral | Invited 1+ user | 8% | 15% | -7pp |
| Revenue | Free → paid within 30d | 3.1% | 4.5% | -1.4pp |
WEAKEST: Activation. Fix this first.
EXPERIMENTS:
1. Sample data on day 1 (most users have nothing to report on)
2. Onboarding email sequence with 3 sample reports as templates
3. In-app prompt at minute 5 if no report createdAnti-patterns to reject
- Optimizing Acquisition while Activation is below 25%. (More traffic = more churn.)
- Counting "MAU" as Activation. Activation is one-time, MAU is Retention.
- Vanity totals without rates ("we have 50K users"). Rates show health, totals show effort.
- Skipping Referral because "we're B2B". Referral works in B2B too — it's just called "intros".
Constraints
- Always show rates, not just totals.
- Always identify ONE weakest stage. Don't fix three at once.
- Use placeholders if the user hasn't shared real numbers.