The AARRR funnel is a framework for measuring five parts of the customer lifecycle: Acquisition, Activation, Retention, Referral and Revenue. It helps a team turn “grow faster” into specific questions: where people arrive, whether they get value, whether they return, whether they bring others and whether the relationship generates revenue.
Use it to diagnose growth, then choose metrics that reflect your product and business model. A dashboard full of signups is of little help if nobody agrees on what an activated or retained customer does.
Dave McClure and the origin of pirate metrics
Dave McClure set out the model in his 2007 presentation Startup Metrics for Pirates. The deck hosted under his own SlideShare account lists Acquisition, Activation, Retention, Referral and Revenue. Its publication metadata dates it to 2007.
The nickname pirate metrics is a mnemonic: say the initials “AARRR!” aloud and they resemble a pirate's cry. The presentation uses that joke to make the five categories memorable. Its practical emphasis is on a small set of actionable conversion metrics, rather than a large collection of activity counts.
What the AARRR funnel measures
The stage definitions below follow McClure's model. The metrics and mistakes are suggested applications, not benchmarks or universal event definitions.
Stage | Question it answers | Example metric | Common mistake |
|---|---|---|---|
Acquisition | Where do new users or customers come from? | Qualified trial starts by channel | Treating all traffic as equally valuable |
Activation | Do they experience initial value? | Accounts completing the core workflow / eligible new accounts | Calling registration activation without checking product use |
Retention | Do they return and continue receiving value? | Accounts repeating the core action in a defined period / eligible cohort | Mixing cohorts or counting any login as useful activity |
Referral | Do existing users bring in new users? | Referred accounts that activate / eligible active accounts | Counting invitations without following the people invited |
Revenue | Does the relationship generate income? | Trial-to-paid conversion for a defined cohort | Celebrating first sales without examining margin or retention |
A pageview can be an acquisition event. It does not by itself establish that the visitor is qualified. Likewise, an account creation can be the start of the measurable journey without being the first value moment. Define the behavior behind each label before building charts.
1. Acquisition: measure qualified entry, not just traffic
Acquisition identifies how people enter the journey. Search, paid media, email, affiliates, communities and referrals are possible channels. Choose an entry event you can measure consistently, such as a qualified visit, a trial start or a demo request.
Keep volume, cost and downstream quality together. For each channel, review acquired users or accounts, cost per acquisition, activation and eventual paid conversion. Define customer acquisition cost separately from cost per signup: their denominators are different.
For a B2B product, compare trial starts with activated accounts and paying customers. For a publisher, choose a useful reading behavior rather than automatically copying the software metric. These are modeling choices; validate them against your own economics.
2. Activation: define the first real value moment
Activation describes an initial useful experience. Registration is only a candidate event, not proof that someone received value. Amplitude's application of AARRR also separates acquisition from the experience that makes users want to continue.
Possible events include connecting a data source and viewing a report, publishing a first marketplace listing, or completing a shared task with a teammate. Treat each as a hypothesis about value, not a rule for every product.
A simple activation test
Ask:
- What useful result does this action represent for the customer?
- Can it occur early enough for onboarding to influence it?
- Do comparable users who complete it return or pay more often than those who do not?
Compare users from the same acquisition period and relevant segment. An observed association with retention can help select an event; it does not prove that making people perform the action causes retention.
For reporting, specify the activation event, eligible population and completion window. An activation rate can be defined as activated accounts divided by eligible new accounts within that window. Track time to activation alongside the rate to locate delays in the experience.
3. Retention: the stage that changes the economics
Retention measures whether customers continue to use, buy or engage with the product over a meaningful period. A cohort is a group that shares a starting point, such as the month of acquisition or activation.
Reforge's retention analysis explains how retention connects to acquisition, monetization and referral opportunities. For subscriptions in particular, continued payment changes the revenue generated by a customer cohort.
Choose a return action and time window that match the product's use. A collaboration tool and an occasional travel purchase need different definitions. Useful measures can include repeat core actions, repeat purchases, renewals, customer churn or revenue retention.
For a cohort retention rate, divide cohort members who perform the defined return action in the target period by the eligible starting cohort. Keep the starting point and denominator fixed. Compare cohorts at the same age, and mark recent cohorts that have not yet reached the observation window.
4. Referral: distinguish advocacy from a real growth mechanism
Referral measures whether existing users bring new users into the system. Invitations, shared documents, referral links and recommendations are possible mechanisms.
Track the outcome after the invitation: acceptance, signup, activation and later retention or payment. An invitation count is useful operationally, but it does not tell you whether the referred person becomes a valuable participant.
If using activated referrals per active account, define both the referring population and the attribution window. Separate the act of referring from the acquisition channel of the person referred so the same relationship can be analyzed from both sides.
5. Revenue: measure economic quality, not only conversion
Revenue describes monetization. Depending on the business, track paid conversion, average order value, revenue per account, recurring revenue or expansion revenue. Analyze margin, acquisition cost and payback alongside these revenue measures rather than treating them all as interchangeable revenue metrics.
Compare paid conversion within a defined cohort. Then examine whether customers remain, buy again or expand. First payment and continued economic contribution answer different questions.
Use billing or commerce records to establish realized revenue. Document how refunds, cancellations and discounts affect the reported value.
Why referral comes before revenue in the original order
McClure's 2007 slides put Referral before Revenue. If you encounter acquisition, activation, retention, revenue, referral elsewhere, the final two categories have been swapped.
Keep the original order when explaining the acronym, but model the actual journey separately. The slides include referral paths back into acquisition; the categories are not a requirement that every customer must refer someone before paying. Reforge's growth-loop framework provides a complementary way to describe outputs that feed another cycle of growth.
How to build your AARRR model
For each stage, write down:
- Event: the exact behavior or business outcome.
- Entity: user, account, customer or transaction.
- Denominator: the population eligible to complete the event.
- Time window: when completion or return is measured.
- Source of truth: product analytics, CRM, billing or commerce.
- Decision and owner: what the metric will change and who acts on it.
For an illustrative SaaS model, use qualified trial starts for acquisition, completion of the core workflow for activation, repetition of that workflow in a later defined period for retention, activated invited accounts for referral and the first paid subscription for revenue. This is a suggested starting model, not an industry standard.
Keep user and account measures distinct. If several people share one paying account, user activation and account conversion need separate denominators. Document identity rules so one customer is not counted as several unrelated customers across systems.
Diagnose the constraint before adding more acquisition
Start by locating an economically important drop. If trial starts increase while the share reaching the core workflow falls, investigate acquisition quality and onboarding together. If activation is stable but repeat use declines, inspect the retention definition and product experience before buying more traffic.
Choose an intervention, state the expected downstream effect and compare compatible cohorts after the change. Prioritize the likely contribution to business value, the evidence and the cost of the intervention. The largest percentage drop is not automatically the best opportunity.
Segment the funnel before trusting averages
Break out acquisition channel, customer segment, plan, use case, device or acquisition cohort when those differences affect a decision. Compare customers at the same maturity rather than a recent cohort with one that has had much longer to renew or pay.
A lower cost per signup is not enough to choose a channel. Compare activation, retention and revenue for the acquired cohort. Preserve the definitions across segments so differences reflect customer behavior rather than different counting rules.
AARRR is a diagnostic framework, not the entire growth system
AARRR helps locate a weak part of the lifecycle. It does not explain the cause by itself. Pair the numbers with investigation of the customer experience and the economics of the proposed change.
Reforge's growth-loop analysis focuses on how one cycle produces inputs for the next. Use AARRR to inspect lifecycle performance, loops to describe reinvestment and unit economics to assess whether the resulting growth can pay for itself.
The operating checklist
Before using the model to make decisions, verify that:
- each stage has an explicit event, entity, denominator and window;
- activation represents a useful experience you can investigate;
- retention compares cohorts at the same age;
- referral follows the people referred through activation;
- revenue reconciles with the business system;
- channel comparisons include downstream quality;
- each metric has an owner and a decision attached to it.
Start with a small set of primary metrics. Add detail when it resolves a concrete question, and record changes to definitions so later comparisons remain interpretable.
Sources
All accessed on October 11, 2026.
- Startup Metrics for Pirates, SlideShare. Accessed October 11, 2026.
- AARRR: Come Aboard the Pirate Metrics Framework, Amplitude. Accessed October 11, 2026.
- The One Growth Metric that Moves Acquisition, Monetization, and Virality, Reforge. Accessed October 11, 2026.
- Growth Loops are the New Funnels, Reforge. Accessed October 11, 2026.
