Pirate
Metrics (AARRR).
A practitioner-written reference on AARRR: Dave McClure's 5-stage growth framework, its history, the criticisms from the growth-loops camp, and how modern growth orgs use it as a diagnostic checklist alongside loops as the strategic frame.
By Daria Dovzhikova · Updated May 2026
TL;DR
- AARRR is Dave McClure's 5-stage growth framework — Acquisition, Activation, Retention, Revenue, Referral — introduced at the 2007 Startonomics conference.
- Its weakness is the linear funnel framing; growth loops describe the compounding systems AARRR can't, which is why modern orgs run both.
- Treat AARRR as a diagnostic checklist for whether your current cohort is healthy, not as the growth strategy itself.
What AARRR is
Pirate Metrics (AARRR) is Dave McClure's 5-stage growth framework: Acquisition, Activation, Retention, Revenue, Referral. McClure introduced it at the 2007 Startonomics conference; it became the default growth-analytics frame for the 2010s startup generation. The name comes from the mnemonic sounding like a pirate; the categories stuck because they forced startups to instrument the full customer journey instead of optimizing vanity metrics.
The framework's critique came from Brian Balfour's 2018 essay at Reforge: real growth motions compound through loops, not funnels. AARRR describes linear movement; loops describe self-reinforcing systems. See the growth loops reference for the alternative frame. Modern growth orgs use both.
The five stages
Acquisition
How users find you. Channels: organic search, paid, social, referral, content, partnerships. Measure: cost per acquisition (CPA), visitor-to-signup conversion rate.
Activation
How quickly new users hit the aha moment. Onboarding completion, time-to-first-value, key feature usage. Measure: % of signups who complete a defined activation milestone within their first session.
Retention
How long users stay engaged. Day-1, day-7, day-30 return rates. Cohort retention curves. Measure: cohort retention at multiple time horizons; flat curves indicate product-market fit.
Revenue
How users monetize. Free-to-paid conversion, expansion revenue, ARPU. Measure: LTV, gross margin, payback period. See /tools/ltv-cac-calculator for the standard math.
Referral
How users bring more users. Viral coefficient, referral rate, NPS. Measure: net new users per existing user; >1.0 means the loop is self-sustaining.
The five stages, side by side
Definition, primary metric, a worked DevTools example, the common failure mode, and the role that owns it. Five rows that make the framework auditable instead of decorative.
| Stage | Definition | Primary metric | DevTools example | Failure mode | Owner |
|---|---|---|---|---|---|
| Acquisition | How users first discover the product | Cost per acquisition (CPA), visitor-to-signup conversion | GitHub README → docs landing page → free-tier signup | Optimizing volume on a channel that doesn't convert to activation | Growth marketing + DevRel (for community-driven channels) |
| Activation | How quickly new users reach the aha moment | % of signups completing a defined activation milestone in first session | First successful API call, first deploy, first metric ingested | Quickstart that takes longer than 5 minutes; tool dies before activation | Product + DevRel (for onboarding flow, docs, quickstart) |
| Retention | How long users stay engaged after activation | Day-1 / Day-7 / Day-30 cohort retention; flat curves = PMF | Weekly active developers per workspace, recurring API calls per account | Treating retention as a marketing problem instead of a product problem | Product (primary), with growth supporting re-engagement |
| Referral | How existing users bring new users | Viral coefficient (new users per existing user), referral rate, NPS | Workspace invites, public dashboards, embed CTAs, OSS dependency mentions | Adding a referral program before the product is referable | Product + growth (for in-product loops), DevRel (for community advocacy) |
| Revenue | How users convert to paying and expand over time | Free-to-paid conversion, expansion revenue, ARPU, LTV, payback period | Free tier → team plan → enterprise contract; usage-based meter crossing paid threshold | Pricing page that requires a demo to see real numbers | Pricing + sales (for enterprise), product (for self-serve monetization) |
Stage definitions follow Dave McClure's original 2007 Startonomics framing; failure modes and owner roles are drawn from 12 years of running growth diagnostics inside developer-first companies including JetBrains, Lightrun, and Odigos.
By the numbers
For developer-first products, the back half of the AARRR funnel is where most growth quietly leaks. One data point that frames the retention bar it has to clear — and a reminder that the funnel diagnoses your current cohort while loops diagnose whether the business compounds.
Net dollar retention separating top-quartile SaaS from the rest per OpenView's benchmarks — the Retention and Revenue stages of AARRR made concrete. A pirate-metrics dashboard that tracks acquisition but not this number is measuring the wrong end of the funnel.
OpenView SaaS Benchmarks · 2024→Each figure links to the primary source. If a number moves in a subsequent annual report, this page gets updated.
FAQ
What are Pirate Metrics?
Pirate Metrics, also called AARRR, is Dave McClure's 5-stage growth framework: Acquisition, Activation, Retention, Revenue, Referral. The name comes from the AARRR mnemonic sounding like a pirate. McClure introduced it at the 2007 Startonomics conference; it became the default growth-analytics frame for the 2010s startup generation.
Are pirate metrics still relevant?
Partially. The 5 categories are still useful as a checklist for growth diagnostics, but the strict funnel framing has been challenged by growth-loop frameworks (Brian Balfour at Reforge, Andrew Chen on network effects). Funnels describe linear movement through stages; loops describe compounding systems. Modern growth orgs use both: AARRR as a diagnostic checklist, loops as the strategic frame. See /growth-loops for the loops alternative.
What's the criticism of AARRR?
Three main criticisms. (1) Linear: real growth motions compound through loops, not funnels. (2) Order-dependent: the framework implies acquisition → activation → retention → revenue → referral, but for many businesses (notably PLG SaaS), retention precedes revenue precedes referral, and acquisition can be downstream of referral. (3) Inputs vs outputs: the categories mix inputs (acquisition) and outputs (revenue), which makes the framework hard to operationalize as a single dashboard. Despite these critiques, AARRR remains useful as a diagnostic checklist.
Who created AARRR Pirate Metrics?
Dave McClure, then-founder of 500 Startups, introduced AARRR at the 2007 Startonomics conference. His original slide deck is widely cited; it framed startup growth measurement around the 5 categories as a counterpoint to vanity metrics like total registered users. The mnemonic stuck because it was memorable; the categories stuck because they were useful.
How is AARRR different from growth loops?
AARRR describes a customer's linear journey through 5 stages. Growth loops describe self-reinforcing systems where the output of one stage feeds back as input to the next iteration of the same loop. AARRR asks 'where are we losing people?'; loops ask 'what compounds?'. Modern growth thinking uses both: loops for strategy (what scales), AARRR for tactics (where is friction in the current cohort). See /growth-loops for the loops framing.
Should startups still use AARRR?
Use it as a diagnostic checklist, not as the strategic frame. The 5 categories force you to instrument each stage of the customer journey, which is useful early-stage hygiene. But don't treat AARRR as the growth strategy itself; the strategy is in the loops, the positioning, and the channel mix. AARRR tells you whether your current cohort is healthy; loops tell you whether the business will compound.
Diagnosing your funnel?
Try the funnel analyzer.
Multi-stage funnel analyzer surfaces the biggest single drop-off across your AARRR stages. Free, no signup.
Open the funnel analyzerReady when you are.
Discovery calls are 20 minutes. First one's on me.