AARRR Pirate Metrics: Dave McClure's Growth Funnel Explained

By Harry Nguyen · Updated Oct 12, 2026

AARRR, or pirate metrics, is a growth framework from investor Dave McClure that sorts a customer's journey into five stages: acquisition, activation, retention, referral and revenue. You pick one metric per stage, measure the drop-off between stages, and fix the weakest link first. The "pirate" nickname comes from how the letters sound when spoken.

Key Facts: AARRR at a Glance

• Dave McClure presented the model in a talk called "Startup Metrics for Pirates," and he published the slides himself on SlideShare under the handle dmc500hats.
• The five stages in McClure's slides are acquisition (users arrive from various channels), activation (a positive first visit), retention (users return), referral (users recommend the product) and revenue (users take a monetizing action). (McClure's slides)
• The conversion rates and dollar values printed on those slides are labelled as examples, not real benchmarks. (McClure's slides)
• McClure has said he developed the model across roles at PayPal, SimplyHired and Mint, starting from a simpler acquisition, engagement and monetization view and adding retention and referral. (Inside, interview translation)
• A later variant, RARRA, reorders the same five stages so retention comes first. It is credited to Gabor Papp and Thomas Petit. (Chameleon)

Where the AARRR Framework Came From

Most startups in the mid-2000s tracked whatever their analytics tool displayed by default: pageviews, downloads, press hits. Those numbers go up and to the right and tell you almost nothing about whether the business works.

McClure wanted a shorter list. In an interview translated by Inside, he described developing the model while working at PayPal, SimplyHired and Mint, looking for a framework founders could use for product and marketing decisions. He said he built on the idea that founders such as Reid Hoffman emphasized acquisition, engagement and monetization, then added retention and referral (Inside). The five stage names, said aloud, give the pirate joke its name.

The framework also became a default vocabulary. Analytics vendors and growth teams all use the same five labels, so a team can say "our activation is weak" and be understood across companies. That shared language is most of its value.

The Five Stages, One at a Time

Each stage asks one question about the customer. McClure's slides pair every stage with example metrics, which you can adapt to your own product (McClure's slides).

Acquisition

Acquisition counts how people first reach you: search, ads, partners, content, outbound. The trap is counting raw volume. Ten thousand visitors who never activate are a cost, not an asset. Measure acquisition by channel, and pair it with cost, so you can see which channels deliver people who go on to later stages.

Activation

Activation is the moment a new user gets a first taste of the value. McClure's slides describe it as a positive first visit and list signals such as a signup or an engaged session. In a product with a trial, it's usually the first completed action that matches your promise: a first report sent, a first invite, a first deal logged. Define it as a behavior, not a page view. The deeper treatment lives in the user activation framework.

Retention

Retention asks whether activated users return. McClure's slide example is a repeat visitor who comes back three or more times in the first 30 days. For subscription software, track the cohort: of everyone who activated in March, how many are still active in April, May and June? A flat line means you've found something people need. A line that slides to zero means more acquisition just fills a leaky bucket.

Referral

Referral measures whether satisfied users bring in others. The slides put a higher value on a referred user who activates than on one who merely visits, which is the right instinct: count the referral that becomes a real user. Referral is the stage most teams skip because it's hard to instrument, and that's a reason to start with something simple, such as invites sent and invites accepted.

Revenue

Revenue asks whether users take a monetizing action, and McClure separates a minimum-revenue user from a break-even user. That distinction matters: a customer who pays $10 and costs $40 to acquire is progress on the funnel and a loss on the business. Once you have expansion revenue, track it alongside first purchase.

AARRR vs Other Growth Frameworks

AARRR is a measurement map. Several neighbors cover different jobs, and mixing them up causes confusion in planning meetings.

A quick rule of thumb: use AARRR to find which stage is weakest, use HEART or the Hook Model to understand why, and use a growth experimentation framework to test the fix.

The Retention-First Variant: RARRA

AARRR is chronological, so it implies a funnel that starts at the top and tries to pour more in. A reordered version, RARRA, was created by Gabor Papp and Thomas Petit to better serve the struggle of retaining mobile app users. It puts retention first and acquisition last, and argues that selling to current customers is easier than finding new ones and that referrals from satisfied users drive growth (Chameleon).

Neither version is wrong. They answer different questions. Use the chronological AARRR view to diagnose a brand-new product with no history. Use the retention-first view once you have an audience and want to spend effort where compounding happens. Many teams run both: AARRR to see the whole path, RARRA to decide the order of work.

How to Run AARRR Step by Step

1. Write down the one-sentence promise. What does the product do for whom? Every stage definition depends on it.
2. Define the activation event as a behavior. Pick the earliest action that predicts a user will stay. If you can't name it, interview five retained users and five churned ones.
3. Choose one metric per stage. Resist the dashboard of 40. One number per stage, each with an owner.
4. Pull a baseline for a single cohort. Take everyone who arrived in a recent four-week window and count how many reached each stage.
5. Find the biggest drop. Compare the percentage that moves from one stage to the next. The weakest transition is your priority, not the stage with the lowest raw count.
6. Form one hypothesis and run one test. For example, "cutting the signup form from seven fields to three will lift activation." Run it with a clear success threshold.
7. Review monthly and re-baseline. After a fix, the weakest stage often moves, and the next bottleneck appears.

Ready-to-Copy AARRR Scorecard

Paste this into a spreadsheet and fill one row per stage for the same cohort. The numbers below are placeholders to show the layout.

Then add a second small table for the transitions between stages:

A worked example, with invented numbers for illustration only: a team acquires 2,000 trial signups in four weeks, 600 activate (30%), 240 are still active in week eight (40% of activated), and 60 pay (25% of retained). The weakest transition is activation, so the next test targets onboarding, not ad spend.

Examples by Business Function

These scenarios are illustrative. Each function reads the same five stages through its own lens.

For the sales-led version, sales funnel stages maps the same flow to deal stages.

Dos and Don'ts

Do

• Pick one metric per stage. A scorecard you can read in a minute gets used.
• Measure by cohort. Averages across all time hide whether the product is getting better.
• Define stages as behavior. "Created a first project" beats "visited the dashboard."
• Fix the biggest transition first. Improving a strong stage is the easiest way to feel productive and the least effective.
• Pair the funnel with a qualitative view. Numbers tell you where users leave; interviews tell you why.

Don't

• Don't treat the funnel as strictly linear. Users skip stages, refer before they pay, and return after months away.
• Don't copy example conversion rates as targets. The figures in McClure's slides are labelled as examples, so your own baseline is the only meaningful starting point.
• Don't count vanity numbers as acquisition. Impressions and downloads are inputs, not outcomes.
• Don't skip referral because it's hard. A rough proxy beats a blank cell.
• Don't expect AARRR to choose your strategy. It diagnoses; it doesn't prescribe a go-to-market motion.

Where AARRR Falls Short

The framework is simple on purpose, and that carries costs. It's built around a single user moving through a product, so it handles B2B buying committees awkwardly: one account may include a champion, an evaluator and a signer, each at a different stage. It doesn't tell you why a stage underperforms, which is why teams add experience and habit frameworks. And the funnel shape can hide loops, such as a product where every new user brings others and acquisition feeds itself.

Treat AARRR as a shared checklist and a prompt for questions, not a model of how growth works.