📈 Analytics & Metrics

AARRR Framework

aarrr · aarrr framework · pirate metrics · pirate funnel · aarrr funnel · acquisition activation retention revenue referral

Definition

AARRR is a framework of five customer lifecycle stages (Acquisition, Activation, Retention, Revenue, Referral), each with its own metric. It is used diagnostically, to answer which stage is capping the growth of the business.

What AARRR is

AARRR is a way of splitting a business into five blocks, each with its own metric and its own toolkit. The framework does not answer the question “what should we do”; it answers the question “where exactly are we losing money right now”.

Stage	The question it asks	E-commerce metrics	Tools
Acquisition	Where do people come from and what does it cost	Visits by channel, share of new users, CAC, ad cost ratio	Paid media, SEO, partnerships, content
Activation	Did the visitor grasp the value on the first visit	Share of sessions with a target action, page depth, add-to-cart rate, bounce rate	Site speed, navigation, search, personalized listings
Retention	Do customers come back	Repeat purchase rate, retention by cohort, interval between orders	Triggered communications, recommendations, assortment, service
Revenue	How much does a customer bring	Average order value, revenue per visitor, purchase frequency, LTV, margin	Cross-sell, upsell, assortment and pricing policy
Referral	Does the customer bring others	Share of orders from referral links, reviews, NPS	Referral mechanics, review programmes, UGC

The value of the framework is not the list of stages as such but the discipline it imposes: every stage has an owner, a metric and a set of hypotheses. Without that, “improving conversion” turns into a pile of disconnected tasks.

Adapting it to an online store

The framework comes from SaaS and mobile products, and two of the five stages change meaning when it is moved into e-commerce.

Activation is not registration. In SaaS, activation is the moment the user gets their first value from the product. In an online store, registration usually happens at checkout, after the purchase decision — as an early signal it is worthless. The working definition of activation in e-commerce is the first meaningful action that separates future buyers from accidental visits:

Candidate activation events:
  — viewing N product pages in one visit
  — using catalogue search
  — applying at least one filter
  — adding to cart or to a wishlist
  — returning to the site within 7 days

The right threshold is found in the data: you look for the action after which the probability of a purchase within 30 days rises sharply. This is the same method used to find the aha moment in product analytics.

Retention is not daily returns. For media and apps, retention is measured by a return on the next day. For an appliance store the normal interval between purchases is months; for groceries it is days. So the retention metric is tied to the natural cycle of the category: the share of customers who made a second purchase within a period equal to the median inter-purchase interval.

ℹ

A useful special case: the second-purchase rate is the most sensitive indicator of the health of a customer base. The step from the first purchase to the second produces the heaviest drop-off in the whole chain in typical projects, and working on it usually pays back faster than working on the fifth purchase.

The criticism: linearity versus reality

The main objection to AARRR is that the framework draws one-way movement while the customer moves differently. They lapse after the third purchase and return six months later for a promotion. They recommend the store to a friend before their own second purchase. They generate revenue and prepare to churn at the same time.

Hence two additions, without which AARRR degrades into a set of averaged numbers:

Cohort analysis. A single retention figure for the whole base hides the dynamics: an influx of new customers masks the decay of retention among older ones. Cohorts by month of first purchase show whether things genuinely got better.
Retention curve. One number — “retention 28%” — does not answer whether the curve flattens out. A flat section means the product has a core of regular customers; a curve tending to zero means the base survives only on new arrivals.

The second limitation is that the framework says nothing about margin. The Revenue stage is easy to improve with discounts while gross profit falls. That is why in e-commerce the Revenue stage is counted in margin, not in turnover.

How to choose the focus metric

The point of AARRR is not to run five dashboards but to pick one bottleneck. The practical procedure:

Count the transitions between stages in absolute numbers for the last quarter, not in percentages. It often turns out that the stage with the worst conversion offers the smallest absolute upside.
Estimate the improvement potential. Lifting retention by 2 percentage points on a base of 300,000 customers almost always outweighs a 5% conversion gain in a narrow segment.
Pick one stage per quarter and one leading metric — that is your north star metric for the period.
Fix the counter-metrics. For average order value growth the counter-metric is the return rate; for conversion growth it is cart margin.
Validate changes with an experiment. A metric moving after a release proves nothing on its own: seasonality and changes in paid media are always sitting right next to it.

Common mistakes

Using AARRR as a report rather than a diagnosis. Five numbers on a dashboard with no chosen bottleneck change no decisions.
Treating registration as activation. This leads to optimising the sign-up form instead of the real barriers in the catalogue and search.
Improving Revenue with discounts. Turnover grows, gross margin falls, and the LTV of a cohort trained on discounts ends up lower.
Measuring retention detached from the category cycle. D7 retention for a furniture store carries no information.
Writing off Referral as unmeasurable. Even basic tagging of referral traffic and tracking the share of orders from referred customers beats having no data at all.

Frequently asked questions

Why is AARRR called pirate metrics?

Because of the acronym: the five stages spell out the growl "AARRR". The name was coined by Dave McClure when he presented the framework for startups. There is no meaning behind it beyond memorability — the substance of the framework is splitting the customer lifecycle into five measurable blocks.

What counts as activation in an online store?

Not registration. In e-commerce, registration often happens at checkout, after the purchase decision has already been made, which makes it useless as an early signal. Activation is usually the first meaningful action that statistically separates future buyers: viewing 3–5 product pages, applying a catalogue filter, using search, adding to cart or to a wishlist. The specific threshold is found in the data, not picked by intuition.

How is AARRR different from a conversion funnel?

A conversion funnel describes the path inside a single visit or a single purchase: visit, product page, cart, order. AARRR describes the whole customer lifecycle, from first contact through repeat purchases and referrals to friends. The funnel sits inside the Acquisition and Revenue stages as their internal mechanics.

What is the main weakness of the framework?

Linearity. AARRR assumes the customer passes through five stages once, in sequence, while in reality they move backwards: they lapse and reactivate, they buy and go inactive again, they refer a friend before their own second purchase. So the framework is used as a map of ownership, while the movement itself is measured with cohort analysis and a retention curve.

Do you need to watch all five metrics at once?

No. The framework exists to find the bottleneck, not to improve five numbers in parallel. In practice a team picks one limiting stage per quarter and one focus metric, and keeps the rest as guardrails — so that an improvement in one place is not paid for by a decline in another.