Pirate Metrics (AARRR)
Pirate Metrics, or AARRR, is a growth framework that breaks the customer lifecycle into five measurable stages: Acquisition, Activation, Retention, Referral, and Revenue.
Key takeaways
- Stages multiply rather than add, so the weakest rate caps everything downstream of it.
- Each stage needs one event definition and one cohort denominator to stay comparable.
- Acquisition reports back in days and retention in months, which skews where teams invest.
- The path is not strictly linear; users lapse, return and refer out of order.
- Committee purchases in enterprise sales map poorly onto per-user activation and referral.
In depth
The framework works because each stage is the denominator for the next. People acquired form the base for activation, the activated form the base for retention, and retained users are the only pool that can realistically refer anyone or keep paying. Since the stages multiply rather than add, the end-to-end rate from first visit to returning customer is the product of five separate conversion rates. That structure is what makes AARRR diagnostic: one weak rate places a ceiling over every number downstream of it.
Each stage responds to different work, so a single initiative rarely moves more than one of them. What skews investment is feedback speed. Acquisition changes are visible within days, activation within a week or two, retention only after months have passed. Teams therefore pour effort into the stage that reports back fastest, even when the ceiling sits further down. Referral is the awkward one: it depends on retention already being solid, so pushing it early usually produces invitations that nobody accepts.
In practice each stage gets one event, one time window and one owner, and the funnel is reviewed against previous cohorts rather than against a target pulled from the air. The stage with the largest relative gap gets the quarter's attention. A scorecard quiz can serve two stages at once: as an entry point it filters for people with the problem you solve, and as a first experience it delivers a result immediately, which is a genuine early success rather than an empty account waiting to be filled.
The model assumes a self-serve product where one person can pass through all five stages. In enterprise sales the buyer, the champion and the daily user are different people, so activation and referral stop mapping onto a single record. The sequence is not strictly linear either: users lapse and return, and some refer before they have retained. Measuring all five stages without a target for any of them produces a dashboard that describes the business but never changes it.
Example in practice
How to measure it
Compute each stage on the same cohort so the ratios chain correctly: activated divided by acquired inside a fixed window, still active at day thirty divided by activated, referrers divided by retained, and payers divided by retained. Multiplying the five gives an end-to-end rate. That product is the number to watch after a change, because it reveals whether an improvement upstream survived or was cancelled out further down the funnel.
Read the funnel by acquisition source as well as by cohort. Two channels can deliver identical volume and differ several times over at activation, and a blended figure conceals that completely. Compare each stage rate against the same stage in an older cohort so product changes and seasonality are visible. Absolute counts on their own cannot distinguish a genuinely better funnel from a temporarily larger top of it.
Common mistakes
The most damaging error is defining activation as signup or first login. That rate looks healthy while almost nobody reaches value, and the real leak stays invisible until retention numbers arrive months later. Choose an event that actually predicts whether someone is still around in eight weeks, state the time window in the definition, and revisit it whenever the product changes, since the old aha moment can quietly stop being the one that matters.
The second is measuring all five stages faithfully and acting on none of them. The weekly funnel review becomes a ritual where every number is noted and nothing is owned. Pick the one stage with the widest gap against a comparable earlier cohort, set a single target for the quarter, and treat the other four as guardrails that must not degrade. A framework used purely for reporting has stopped being a growth tool.
Frequently asked questions
What does each letter in AARRR stand for?
AARRR stands for Acquisition, Activation, Retention, Referral, and Revenue. Each represents a distinct stage of the customer journey that you can measure and optimize separately.
Is AARRR still relevant for modern SaaS?
Yes, it remains a clear, durable mental model for diagnosing funnel health. Many teams now pair it with a North Star Metric to keep the five stages aligned to one overarching goal.
Which AARRR stage should you work on first?
The one with the widest gap relative to comparable cohorts, not the one with the smallest absolute number. A stage can look poor simply because it sits late in a multiplying chain. Calculate each conversion rate, compare it to your own earlier cohorts, and pick the biggest relative drop. Fixing an upstream leak also inflates every stage after it, which is worth accounting for.
How do you define an activation event properly?
Find the action that best separates users who are still around weeks later from those who are not, then attach a time window to it. Look for a behaviour that requires real effort and delivers real value, such as completing a first piece of work or connecting a data source. Vague events inflate the rate. Re-derive the definition whenever the product's core workflow changes.
Is AARRR still useful for B2B SaaS?
It works well for self-serve and product-led motions and awkwardly for committee-driven enterprise deals, where the buyer, champion and user are different people. In that case map the stages to the account rather than the individual: acquisition becomes an engaged account, activation becomes a working deployment, referral becomes an internal expansion. The logic survives, the unit of measurement changes.
What is the difference between AARRR and RARRA?
RARRA reorders the same stages to put retention first, arguing that acquiring users into a product that does not hold them wastes the spend. The stages themselves are unchanged. Treat the difference as a statement about sequencing priorities rather than a competing model: AARRR describes the customer's path, RARRA describes the order in which a team should fix it.
How does AARRR relate to a North Star Metric?
AARRR is the diagnostic layer beneath a North Star Metric. The North Star is one number expressing delivered value, while the five stages explain which part of the journey moved it. Teams usually keep the North Star for alignment and the AARRR rates for troubleshooting, because a single headline number tells you something changed but not where.
How often should the AARRR funnel be reviewed?
Monthly for the full set, weekly for the one stage currently being worked on. Retention and referral move too slowly to justify weekly interpretation, and reading noise as signal leads to reversing decisions before they have had time to show an effect. Keep cohort charts rather than period snapshots so slow-moving stages remain readable across reviews.