Growth teams rarely suffer from a shortage of metrics. The harder problem is deciding which numbers describe the business and which merely describe activity.
The AARRR framework gives teams a compact way to organize that decision. Introduced by investor and entrepreneur Dave McClure, the model groups the customer lifecycle into five stages: Acquisition, Activation, Retention, Referral and Revenue. It is often called “Pirate Metrics” because the acronym sounds like “AARRR.”
The framework is useful because it forces a company to translate a vague objective such as “grow faster” into a set of observable customer behaviors. But AARRR only becomes valuable when each stage is defined around the economics and behavior of the specific business. Copying generic metrics into a dashboard does not create a growth model.
This guide explains how to build an AARRR funnel that can actually be used to diagnose growth.
What the AARRR funnel measures
At a high level, the five stages answer five different questions:
- Acquisition — Where do new users or customers come from?
- Activation — Do they experience meaningful value?
- Retention — Do they continue to receive value over time?
- Referral — Do existing users help create new users?
- Revenue — Does the relationship generate sustainable economic value?
The important word is meaningful. A pageview may be an acquisition event, but it may tell you very little about whether the person is a serious prospect. A signup may look like activation, but it is not activation if users routinely create accounts and never reach the product’s core value.
A strong AARRR model therefore begins with event definitions, not charts.
1. Acquisition: measure qualified entry, not just traffic
Acquisition identifies how people first enter the measurable customer journey.
Typical acquisition sources include:
- organic search;
- paid search;
- paid social;
- affiliates;
- creator partnerships;
- referrals;
- direct traffic;
- email;
- communities;
- product-led distribution.
The mistake is treating all incoming traffic as equivalent.
For a B2B software company, “website session” may be too weak to serve as the main acquisition event. A better event might be a qualified demo request, trial start or account creation. For an ad-supported publisher, the economics are different: a high-quality session may itself be economically important.
Acquisition should therefore be measured by both volume and quality.
Useful metrics include:
- new qualified visitors or users by source;
- cost per acquired user;
- customer acquisition cost by channel;
- visitor-to-signup rate;
- signup-to-qualified-lead rate;
- contribution of each channel to activated or paying customers.
A channel that produces cheap signups but weak retention can be worse than a more expensive channel that consistently produces valuable customers.
2. Activation: define the first real value moment
Activation is often the most poorly defined stage.
A registration is easy to track, so teams frequently call it activation. But registration describes what the user did for the company. Activation should describe whether the user experienced enough value to justify continuing.
The right activation event varies by product.
Examples might include:
- an analytics customer connecting a first data source and viewing a useful report;
- an ecommerce customer completing a first purchase;
- a marketplace seller publishing a first listing;
- a collaboration product user inviting a teammate and completing a shared task;
- a publisher subscriber reading multiple articles after subscribing.
The best activation metric is usually a behavior that is early, observable and correlated with later retention.
That correlation matters. Teams should compare cohorts of users who completed candidate activation events against users who did not. If an event has no meaningful relationship with downstream retention, it may be convenient to measure but strategically unimportant.
A simple activation test
Ask three questions:
- Does this action indicate the user has received a meaningful benefit?
- Does it happen early enough to influence?
- Are users who complete it materially more likely to retain or pay?
If the answer to the third question is unknown, the next task is analysis, not optimization.
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.
This is where AARRR becomes more than a marketing funnel.
Reforge has argued that retention affects several other parts of the growth model. Better retention can increase lifetime value, improve the economics of paid acquisition and create more opportunities for referral or user-generated distribution. In other words, retention does not merely reduce losses; it can change how much a company can afford to spend on growth.
Retention windows should reflect natural product frequency.
A daily collaboration tool may care about weekly active retention. A payroll platform may be used several times per month. A travel product might have long gaps between transactions but still maintain a valuable relationship.
Common retention metrics include:
- day, week or month retention;
- repeat purchase rate;
- renewal rate;
- logo retention;
- revenue retention;
- churn;
- cohort retention curves;
- frequency of core actions.
Avoid comparing retention metrics across unrelated products without context. A “good” retention rate depends heavily on category, usage frequency, contract structure and customer segment.
4. Referral: distinguish advocacy from a real growth mechanism
Referral measures the degree to which existing users bring new users into the system.
It can happen intentionally through referral programs or organically through product use.
Examples include:
- inviting colleagues;
- sharing a public document;
- referring a friend;
- publishing user-generated content that attracts search traffic;
- sending a product-generated link to a client;
- word-of-mouth recommendations.
The most useful referral metrics are not simply shares or invitations. They track what happens after the referral:
- invitations per active user;
- invite acceptance rate;
- referred signup rate;
- referred activation rate;
- referred customer conversion;
- retention of referred users;
- viral coefficient, where relevant.
A referral loop is only valuable if it creates users who themselves become valuable participants.
5. Revenue: measure economic quality, not only conversion
Revenue closes the economic side of the model.
Depending on the business, revenue metrics may include:
- customer conversion rate;
- average order value;
- average revenue per user;
- monthly or annual recurring revenue;
- gross margin;
- expansion revenue;
- customer lifetime value;
- customer acquisition cost;
- CAC payback period.
Revenue should not be analyzed independently from retention. A business can produce strong initial sales and still have poor economics if customers churn before acquisition costs are recovered.
For subscription products, this is particularly important. A customer who pays for one month is economically different from a customer with the same entry price who retains for two years.
How to build your AARRR model
Start with a blank document rather than your analytics tool.
For each stage, define:
- The event — what exactly happens?
- The entity — user, account, customer, company or transaction?
- The denominator — what population is the rate calculated from?
- The time window — when must it happen?
- The source of truth — product analytics, CRM, billing or another system?
- The business question — what decision will this metric support?
A SaaS model might look like this:
- Acquisition — Primary event: Qualified trial starts · Example metric: Trials by channel
- Activation — Primary event: Core workflow completed · Example metric: Activation rate
- Retention — Primary event: Core workflow repeated in week four · Example metric: Week-4 retained accounts
- Referral — Primary event: Invited account activates · Example metric: Activated referrals per retained account
- Revenue — Primary event: Account becomes paying · Example metric: Trial-to-paid rate, CAC, LTV
The exact events matter more than the labels.
Diagnose the constraint before adding more acquisition
One of the most useful applications of AARRR is identifying where additional acquisition would be wasted.
Imagine 100,000 monthly visitors produce:
- 5,000 signups;
- 1,500 activated users;
- 450 retained users;
- 180 paying customers.
A team focused only on traffic might spend aggressively to move 100,000 visits to 150,000. But if activation or retention is the binding constraint, much of the additional traffic will simply leak out later.
A better operating sequence is:
- locate the largest economically important drop;
- understand why it occurs;
- improve that stage;
- verify the downstream effect;
- only then decide whether additional acquisition deserves more budget.
The largest percentage drop is not automatically the priority. The priority is the change with the greatest expected effect on business value.
Segment the funnel before trusting averages
Company-wide AARRR metrics can hide important differences.
Segment by variables such as:
- acquisition channel;
- geography;
- device;
- customer size;
- plan;
- use case;
- first product adopted;
- cohort month;
- campaign.
Suppose paid social customers have a lower CAC than search customers. That looks attractive until cohort analysis shows search customers retain twice as long. The average funnel obscures that tradeoff.
AARRR becomes more useful when the team can compare the economics of different customer paths.
AARRR is a diagnostic framework, not the entire growth system
Funnels have an important limitation: they are usually represented as one-directional.
Reforge’s growth-loop framework highlights what this misses. In many businesses, retained customers create content, referrals, usage signals or revenue that feed the next cycle of acquisition. Growth can therefore be circular and compounding rather than simply a sequence from top to bottom.
That does not make AARRR obsolete.
A useful way to combine the two models is:
- use AARRR to diagnose lifecycle performance;
- use growth loops to understand how outputs feed future inputs;
- use unit economics to decide whether growth is sustainable.
The funnel tells you where the journey is weak. The loop explains how the system can compound.
The operating checklist
Before calling an AARRR model complete, verify that:
- every stage has a precise event definition;
- activation represents customer value, not administrative activity;
- retention uses a time window appropriate to the product;
- referral measures downstream quality, not only sharing;
- revenue includes the economics of retention and acquisition;
- metrics can be segmented by channel and customer cohort;
- the team knows which system is the source of truth;
- each metric connects to a decision the team can make.
AARRR is most useful when it reduces complexity. If the “simple” growth funnel contains dozens of primary metrics and nobody agrees on the event definitions, the model has failed its purpose.
The goal is not to make the customer journey fit five letters. It is to create a shared operating model that tells the team where growth is working, where it is leaking and what deserves attention next.
