AARRR Metrics Framework: A Founder's Complete Guide
AARRR, Dave McClure's "Pirate Metrics," splits your startup into five measurable stages: acquisition, activation, retention, referral, and revenue. Instead of chasing one vanity number, you instrument each stage on its own, find the weakest link, and fix it. This guide walks through what every stage measures and how founders act on it.
Quick Answer: AARRR (Pirate Metrics) tracks five startup stages: acquisition (how users find you), activation (their first success), retention (whether they return), referral (whether they tell others), and revenue (whether they pay). Diagnose your weakest stage first, then improve one stage at a time.
Why five stages beat one big conversion number
A single conversion number hides where growth actually breaks. If "sign-ups to paying customers" is 2%, that one figure cannot tell you whether the problem is bad traffic, a confusing first session, users who never return, or a price nobody will pay. Those are four different fixes, and one number points at none of them.
AARRR solves this by treating your product as a sequence of hand-offs rather than a single leap. Each stage passes users to the next, and each has its own drop-off you can see and improve in isolation. When you know that activation is fine but retention collapses in week two, you know exactly where to spend the next sprint.
Dave McClure introduced the framework in 2007 at 500 Startups (then 500 Hashtags) in a talk titled "Startup Metrics for Pirates," so named because the five initials spell AARRR. His argument was blunt: most founders drown in dashboards full of numbers that feel productive but never change a decision. Pick the metric that matters for the stage you are on, he said, and ignore the rest.
The stages also compound. A 10% gain at each of five stages multiplies, not adds, so the same effort spread across the funnel outperforms one heroic push at the top. This is the core insight behind funnel math and how startup conversion rates stack: improving the worst stage usually returns more than improving the stage you already understand best.
Consider a concrete diagnosis. Two founders both see weak paying-customer numbers. The first discovers strong acquisition and activation but a retention curve that falls off a cliff in week two, so their fix is the product experience that fails to bring people back. The second finds healthy retention among the few who activate, but most sign-ups never reach first value at all, so their fix is onboarding. Same symptom at the bottom, opposite problems, and only a staged view separates them.
For early-stage founders, the framework doubles as a diagnostic. Before product-market fit, you are not trying to grow, you are trying to locate the exact stage where interest turns into indifference, and then decide whether the product, the message, or the market needs to change.
The five AARRR stages at a glance
Here is the whole framework in one view: what each stage asks, what you instrument, and the kind of signal that tells you it is working. The table is qualitative on purpose, because healthy numbers differ wildly between a daily consumer app and an annual B2B contract.
| Stage | The question it answers | What founders instrument | A healthy signal looks like |
|---|---|---|---|
| Acquisition | How do people find us? | Traffic and sign-ups broken out by channel | A channel that reliably sends genuinely interested visitors |
| Activation | Do new users reach first value? | A defined "aha" event early in onboarding | A rising share of new users who hit that event |
| Retention | Do users keep coming back? | Cohort return behavior over time | A retention curve that flattens instead of falling to zero |
| Referral | Do users bring others? | Invites, shares, and word-of-mouth sign-ups | New users arriving because an existing user told them |
| Revenue | Will users pay? | Conversion to paid, expansion, willingness to pay | Customers paying without heavy discounting or arm-twisting |
The takeaway: no single stage is "the metric." Each answers a different question, and your job is to find which one is currently leaking, not to memorize a target for all five.
A note on ordering. The acronym reflects McClure's original sequence: Acquisition, Activation, Retention, Referral, Revenue. Many teams later reordered it into a strict top-to-bottom funnel and moved Revenue earlier, so you will see the excerpt above and plenty of other guides list revenue before referral. Both are common; the letters matter more than the exact position, and this guide keeps McClure's original order while flagging where the funnel view differs.
Acquisition: how strangers first discover your product
Acquisition measures how people find you and which sources send the ones worth having. It is the top of the funnel, and the temptation is to reduce it to a single traffic count. The more useful move is to break acquisition out by channel, so you can see not just how many people arrive but where the good ones come from.
Instrument acquisition by source, not in aggregate. Track visits, sign-ups, and cost separately for each channel: organic search, content, paid ads, communities, referrals from partners, direct. A blended "1,000 visitors" number is nearly useless; "400 from a niche community that convert well, 600 from paid that bounce" tells you where to double down.
The metrics that matter here are qualitative comparisons more than absolute figures:
- Channel volume — which sources actually send people, and at what relative scale
- Channel quality — which sources send visitors who go on to activate, not just click
- Channel cost — what you spend, in money or time, to earn one qualified visitor from each
The classic founder trap is optimizing acquisition first. Pouring traffic, especially paid traffic, into a product that does not yet activate or retain is the "leaky bucket" problem: you pay to fill the top while users drain out the bottom. Acquisition is worth scaling only once the stages beneath it hold water.
Judge channels on what they produce downstream, not on what they produce at the top. A channel that sends cheap sign-ups who never activate is more expensive than a pricier channel whose users stick, once you trace each source through to retention and revenue. This is why attribution belongs to acquisition: you need to know not just that a visitor arrived, but which channel eventually yielded a durable, paying user.
One more distinction worth drawing early: acquisition is about strangers becoming visitors or sign-ups, not about them becoming happy or paying. Conflating "we got a lot of sign-ups" with "the product is working" is how teams celebrate a metric that means far less than it feels like it should.
Activation: the first moment a user feels real value
Activation measures whether a new user reaches their first genuine success, the "aha" moment where the product's promise becomes real for them. Acquisition gets someone in the door; activation is the first time they actually experience why they came. It is arguably the highest-leverage stage for early products, because everything downstream depends on it.
The core task is defining your activation event. This is a single, specific action that reliably predicts a user will stick around: sending a first message, inviting a teammate, completing a project, connecting a data source. Vague definitions like "logged in" or "clicked around" do not count, because they do not separate users who found value from those who bounced.
A few principles for choosing that event well:
- It should correlate with retention. Look at which early action best predicts whether a cohort comes back, then treat that as your activation bar.
- It should be reachable in the first session. If first value takes a week, most users never see it. Shorten the path.
- It should be one thing, not five. A single, clear milestone is easier to instrument, communicate, and improve than a fuzzy "engaged" score.
Once defined, activation becomes a lever you can pull directly. Reducing time-to-value, cutting onboarding steps, adding a guided first-run experience, or pre-filling example data all move the share of new users who activate. Because activation is a leading indicator of retention, improving it tends to lift the stages beneath it for free.
Activation is also where validation and metrics meet. If you have run a complete startup idea validation process before building, you already have a hypothesis about what "first value" should be; activation is where you confirm the product delivers it in practice, not just in a landing-page promise.
Retention: whether users actually keep coming back
Retention measures whether users return over time, and it is the stage most founders underrate. Acquisition and activation get someone to try the product once; retention is the honest test of whether it earned a place in their life or work. Many practitioners treat retention as the truest proxy for product-market fit, because a product people keep using is, by definition, one they value.
Measure retention with cohorts, not a single average. Group users by the week or month they signed up, then track what fraction of each cohort is still active in later periods. A blended "monthly active users" figure can rise even as every individual cohort bleeds out, because new sign-ups mask the churn underneath. Cohorts expose the leak; aggregates hide it.
The shape you are looking for is a retention curve that flattens. Every cohort declines at first, that is normal, but a healthy product's curve levels off at some stable plateau, meaning a durable core keeps returning. A curve that slides toward zero says users try the product and abandon it, no matter how good acquisition looks.
What counts as "returning" depends entirely on the product's natural cadence:
- A daily-use tool (messaging, tracking, news) should see users back most days
- A weekly workflow tool might be healthy at once or twice a week
- A quarterly or seasonal product (tax, planning, travel) should not be judged on daily numbers at all
Because retention feeds every stage below it, it is often the right anchor for your top-line metric. Retention behavior is exactly the kind of durable engagement that a well-chosen North Star metric for founders is meant to capture, tying the whole team to whether users keep getting value rather than to a one-time sign-up spike.
Referral: turning satisfied users into a growth channel
Referral measures whether existing users bring new ones, converting your happiest customers into an acquisition channel that costs little and trusts well. When it works, referral lowers your effective cost of acquisition and creates the compounding loops behind fast-growing products. When it is forced too early, it does nothing.
Referral only works downstream of retention. Users who have not yet found lasting value have nothing genuine to recommend, and incentivizing them to invite friends just imports low-quality sign-ups who churn. The sequence matters: earn retention first, then ask users to spread the word.
There are two broad flavors worth distinguishing:
- Organic referral — users tell others because the product is worth talking about, or because it is more useful with collaborators. This is the strongest and cheapest kind.
- Incentivized referral — you offer a reward for invites. It can accelerate a working loop, but it cannot manufacture one that does not exist on its own merits.
To instrument referral, track how many new users arrive attributed to an existing user, whether through invite links, shares, or a simple "how did you hear about us" prompt. Founders often reach for sentiment measures like Net Promoter Score here too, though a stated willingness to recommend is a weaker signal than an actual referral you can count.
The concept some teams borrow is the viral coefficient: how many new users each existing user brings, on average. You do not need a precise figure to act on the idea. Directionally, the question is whether your product's referral loop adds meaningfully to growth or is a rounding error, and whether making the product more collaborative would strengthen it.
Revenue: turning engagement into money
Revenue measures whether users will actually pay, and how much, once they are acquired, activated, and retained. In McClure's original ordering it comes last, not because money matters least, but because sustainable revenue usually follows from the stages before it. A retained, referring user base is far easier to monetize than cold traffic.
Instrument revenue as more than one number. The useful signals include conversion from free or trial to paid, average revenue per user or account, and expansion, whether existing customers grow their spend over time. Together these tell you not just that money arrives but whether the underlying economics can support growth.
The relationship founders return to most is between what a customer is worth and what it costs to get one:
| Concept | What it captures | Why founders watch it |
|---|---|---|
| Willingness to pay | Whether the value is worth real money to users | A product people love but won't pay for has a monetization problem, not a product one |
| Conversion to paid | The share of qualified users who become customers | Reveals whether pricing and packaging match perceived value |
| Lifetime value vs. acquisition cost | What a customer returns against what they cost | The unit economics that decide whether growth is sustainable |
The takeaway: revenue health is a ratio story, not a total-dollars story. A rising revenue number built on customers who churn quickly or cost more to acquire than they ever pay is a warning, not a win.
This is also where the funnel-view and original-order debate has practical stakes. Teams that move Revenue earlier tend to be selling first and building an audience second; teams that keep it last, as here, are betting that retention and referral make monetization far easier when it comes. Neither is wrong, but knowing which bet you are making keeps you from optimizing the wrong stage.
Common mistakes founders make instrumenting AARRR
The framework is simple to draw and easy to misuse. Most failures come not from the model but from instrumenting all five stages before any single one works, or from measuring in ways that flatter the numbers. The table below pairs the most common mistakes with what to do instead.
| Mistake | What it looks like | Do this instead |
|---|---|---|
| Optimizing acquisition first | Scaling ad spend into a product that doesn't yet activate or retain | Fix activation and retention before pouring in traffic |
| Tracking totals, not cohorts | Rising "active users" while every cohort quietly churns | Group users by sign-up period and watch each cohort's curve |
| No defined activation event | "Active" means everything and therefore nothing | Pick one early action that predicts retention |
| Chasing vanity metrics | Big, flattering numbers that never change a decision | Tie every metric to a stage and to an action you'd take |
| Measuring all five at once, early | A cluttered dashboard, no clear priority | Focus on the one leaking stage; ignore the rest for now |
The deepest trap is the vanity metric. A number is a vanity metric if it can only go up and never tells you what to do differently: cumulative sign-ups, total page views, raw download counts. Alistair Croll and Benjamin Yoskovitz make this the spine of Lean Analytics, arguing for "One Metric That Matters" at a time, the single number tied to your current stage, rather than a wall of dials.
Another quiet failure is instrumenting stages you are not ready for. Before product-market fit, referral and revenue optimization are usually premature; the honest work is at activation and retention, where you learn whether anyone actually wants the thing. This is where AARRR shifts from a growth dashboard to a validation tool, and where a platform like Edmired focuses founders on evidence that a stage genuinely works before they scale it.
Finally, watch for measuring the aggregate when the truth lives in the segment. Averages blur the difference between a power-user cohort that loves the product and a long tail that never returns, and it is that difference, not the mean, that tells you where product-market fit is forming.
Key Takeaways
- AARRR breaks a startup into five measurable stages — acquisition, activation, retention, referral, and revenue — so you can find and fix the specific place growth leaks instead of guessing at one blended number.
- Dave McClure created "Pirate Metrics" at 500 Startups in 2007, and the acronym reflects his original order; many teams reorder it into a strict funnel and move revenue earlier, but the letters matter more than the position.
- Activation is the highest-leverage early stage because it is a leading indicator of retention; define one specific "aha" event that predicts users will stick, and shorten the path to it.
- Retention is the truest proxy for product-market fit — measure it with cohorts and look for a curve that flattens, since a blended active-user count can rise while every cohort churns.
- Referral and revenue come after retention works, not before; asking un-retained users to refer or pay imports low-quality sign-ups and masks the real problem.
- Vanity metrics are the core failure mode — if a number can only go up and never changes a decision, replace it with the one metric that matters for your current stage.
- Before product-market fit, AARRR is a diagnostic, not a growth engine — use it to locate where interest turns to indifference, then decide whether product, message, or market needs to change.
Frequently Asked Questions
What does AARRR stand for?
AARRR stands for acquisition, activation, retention, referral, and revenue, the five stages of Dave McClure's "Pirate Metrics" framework. Each stage measures a distinct step in a user's journey, from first discovering the product through paying for it, so founders can track and improve each one separately.
Who created the AARRR framework?
Dave McClure created AARRR in 2007 while at 500 Startups, in a presentation called "Startup Metrics for Pirates." He nicknamed it "Pirate Metrics" because the five initials spell AARRR. His goal was to replace cluttered dashboards with a small set of stage-specific metrics that actually drive decisions.
Is AARRR a funnel or a set of stages?
Both interpretations are common. McClure's original framing is a set of five stages in the order A-A-R-R-R (revenue last), while many modern teams draw it as a strict top-to-bottom funnel and place revenue earlier. Functionally it works as a funnel of hand-offs, since each stage passes users to the next; the ordering label matters less than instrumenting every stage.
Which AARRR stage should I focus on first?
Focus on the earliest stage that is leaking, which for pre-product-market-fit startups is almost always activation or retention. Optimizing acquisition or revenue before users reliably reach first value and return just scales a broken funnel. Find your weakest stage, fix it, then move down.
How is AARRR different from a North Star metric?
AARRR is a full-funnel framework of five stage-specific metrics, while a North Star metric is a single top-line number the whole team rallies around. They are complementary: teams often draw their North Star from the retention or activation stage of AARRR, then use the other stages as the diagnostic inputs that explain why the North Star moves.
Do I need AARRR before product-market fit?
Yes, but as a diagnostic rather than a growth engine. Before product-market fit, use AARRR to pinpoint where users lose interest, usually at activation or retention, instead of optimizing referral and revenue prematurely. It tells you whether the product genuinely works at each stage before you spend money trying to scale it.