AARRR
AARRR, often called "Pirate Metrics" because the acronym reads as "Arrr," is a five-stage metrics framework startups use to track users moving through their product: Acquisition (users arrive), Activation (they have a satisfying first experience), Revenue (they pay), Retention (they come back), and Referral (they bring others). Teams typically build dashboards, growth-team structures, and quarterly goals around these five categories.
Definition
AARRR (Pirate Metrics) is a product analytics framework that organizes user behavior into five sequential categories — Acquisition, Activation, Revenue, Retention, Referral — and treats them as phases of the user's journey.
The five categories
Each letter names a phase the business cares about:
- Acquisition — the user shows up.
- Activation — the user does something the business reads as a first sign of engagement.
- Revenue — the user pays.
- Retention — the user keeps coming back.
- Referral — the user brings other users in.
Teams adopt AARRR as a checklist for what to measure across the user lifecycle, with each category getting its own metric and its own optimization work.
The business sequence, relabeled
The five categories describe what the business wants the user to do, in the order the business wants it to happen. AARRR presents this sequence as a description of the user's experience, but users don't think of themselves as being in the retention phase or generating referrals. They're pursuing their own outcomes, on their own user timeline, with their own reasons for engaging.
AARRR is a provider-biased timeline artifact: the business's preferred sequence of user actions, relabeled as if it described the user's pursuit. The categories aren't wrong because they're incomplete or because some other category is missing — they're the wrong kind of thing to begin with. They measure the user from the perspective of what the company wants next, not from the perspective of what the user is trying to accomplish.
The consequence at the measurement layer: any metric anchored to an AARRR phase inherits the perspective error. An "activation rate" defined as "did the user do the thing that predicts revenue for us" can move without the user making any progress toward what they came for, and can fail to move when users are progressing in ways the metric doesn't see.
Activation, reclaimed
The framework keeps the word "activation" because the industry uses it everywhere and fighting that is a losing battle, but reshapes its content: the user making genuine progress toward the sought state they came for. Operationalized as value milestones — behaviors correlated with the user actually getting what they came for, validated as proxies for real progress rather than for revenue-prediction.
The other four categories aren't reclaimed in the same way. Acquisition, Revenue, Retention, and Referral remain useful as descriptions of business events. The framework treats them as operational categories on the business side, distinct from the user-side transitions it models.
Example
A team running an AARRR dashboard sees activation rate climbing and revenue holding steady; retention is stable. By AARRR's own logic, things are working. But the activation event being counted is "user added a teammate within 7 days" — a behavior the team picked because it correlates with paid conversion. The team can move that number by prompting users to invite teammates during signup. The number goes up. Conversion doesn't follow, because most of the new teammate-adders weren't on track to convert anyway; they were just users complying with a prompt. The metric kept computing exactly what it computed, while what the team thought it was measuring (users making progress toward what they came for) had drifted into fiction. The AARRR framing made the drift hard to see: the dashboard reads as users moving through the phases, but the users aren't doing what the dashboard depicts.