AARRR pirate metrics and metric trees

Dave McClure introduced the AARRR framework, affectionately known as pirate metrics, in his 2007 presentation "Startup Metrics for Pirates." The name stuck because the acronym sounds like a pirate's exclamation, but the framework itself is serious and widely adopted. McClure's core insight was that most startups tracked too many metrics without any organising structure. Revenue was the goal, but the path from a stranger visiting your website to a loyal customer recommending your product involved several distinct transitions, each with its own dynamics, failure modes, and levers.

The framework divides the customer lifecycle into five stages. Acquisition is how users find you. Activation is the moment they experience your product's core value for the first time. Retention measures whether they come back. Revenue captures when they pay. Referral tracks whether they bring others. Each stage is a gate: users either pass through or drop out. By measuring the conversion rate at each gate, teams can identify where the biggest losses occur and focus their efforts accordingly.

The AARRR framework is typically presented as a linear funnel: users flow from Acquisition through Activation, Retention, Revenue, and Referral in a neat sequence. This linear view is useful for a first pass, but it obscures something important. Each stage is not a single metric. It is a cluster of interconnected drivers that influence one another in ways a flat funnel cannot capture. Acquisition quality affects Activation rates. Activation depth affects Retention. Retention strength affects Revenue potential. Revenue per user affects your ability to invest in Acquisition. The stages are not independent compartments. They are branches of a system.

A metric tree makes these connections explicit. Instead of treating each AARRR stage as a single number on a dashboard, you decompose it into the sub-metrics that actually drive it. Acquisition is not just "number of new visitors." It is organic traffic plus paid traffic plus referral traffic, each with its own cost, quality, and intent profile. Activation is not just "percentage who completed onboarding." It is a chain of micro-conversions: sign-up completion, first key action, time to value, and feature discovery. When you decompose each stage, you get a tree where the five AARRR labels become branches, and beneath each branch sit the operational levers your teams control.

This decomposition also reveals cross-stage dependencies that the funnel model hides. Referral, which sits at the bottom of the traditional funnel, actually feeds back into Acquisition at the top. Retention does not just keep users; it directly expands Revenue through upsell and expansion opportunities. A metric tree can represent these feedback loops as shared nodes or cross-branch connections, giving teams a more honest picture of how growth actually works rather than forcing reality into a one-directional pipe.

The real power of combining AARRR with a metric tree emerges when you build the full structure. Start with a top-level outcome metric, typically Revenue or Monthly Recurring Revenue, and decompose downward through the AARRR stages. The tree below shows how a SaaS business might structure this. Revenue sits at the root. Beneath it, the five pirate metrics stages become major branches, each decomposed into the specific drivers that teams can measure and influence.

Notice that the tree is not a strict top-to-bottom funnel. Revenue depends directly on paying customers and ARPU. Paying customers depend on Retention of existing customers and Activation of new ones. New activated users come from Acquisition. Referral feeds back into Acquisition as a channel. This structure reflects how the stages actually interact rather than forcing them into a sequential pipe. Each leaf node is something a team can directly influence: a campaign, a product flow, a pricing page, a support process.

Building this tree forces several conversations that a flat dashboard never triggers. Where does Activation end and Retention begin for your product? Which Acquisition channels produce users that actually activate? Is your Referral loop strong enough to treat as a genuine Acquisition channel, or is it aspirational? These are not abstract questions. They determine where you draw the branch boundaries, which team owns which node, and where you invest your next sprint. The tree structure demands answers because every metric must connect to something above and below it. Orphan metrics that cannot justify their place in the causal chain either need rethinking or removal.

Most teams that adopt pirate metrics implement them as a flat funnel dashboard: five rows, each with a headline metric and a trend line. Acquisition: 12,000 new visitors, up 8%. Activation: 34% onboarding completion, flat. Retention: 62% day-30 return rate, down 3%. Revenue: $84k MRR, up 5%. Referral: 1.2 invites per user, flat. This dashboard is better than nothing. It tells you which stage is weakening. But it cannot tell you why, and it cannot tell you what to do about it.

Consider the Retention line: 62% day-30 retention, down 3%. A flat dashboard raises the alarm but leaves the team guessing. Is retention falling because onboarding quality declined, meaning fewer users reached the aha moment? Is it because a recent product change reduced engagement for a specific user segment? Is it because a competitor launched a feature that pulled away power users? The flat funnel treats Retention as a black box. The metric tree opens the box and shows you which sub-driver changed, so you can respond with precision rather than guesswork.

The other limitation of flat funnels is that they obscure cross-stage interactions. When your Acquisition volume spikes because of a viral campaign, Activation rate often drops because the new cohort is lower-intent than your usual traffic. A flat dashboard shows Acquisition up and Activation down, and the team might mistakenly conclude that the onboarding flow is broken. A metric tree, by connecting Acquisition channel quality to Activation rates, would reveal that the onboarding flow is performing the same as ever for organic users and that the drop is entirely attributable to the campaign cohort. This kind of structural insight is impossible when each stage is tracked in isolation.

One of the most common mistakes with pirate metrics is selecting the wrong metric for each stage. The framework gives you five labels, but it does not prescribe which specific number to track. That choice depends on your business model, your product's user experience, and your current growth stage. A B2B SaaS company and a consumer mobile app will share the same five AARRR labels but track entirely different metrics beneath them. The principles below will help you choose well.

1. 1

   Define the boundary event for each stage

   Before choosing a metric, define the specific user action that marks the transition between stages. Acquisition ends and Activation begins when a user does what, exactly? Signs up? Completes onboarding? Performs a key action? The boundary event must be concrete and measurable. Vague transitions like "the user understands the product" are not measurable and will produce unreliable metrics.

2. 2

   Pick metrics that your team can influence

   Each metric should be something a team can move through deliberate action. "Total revenue" is an outcome, not an actionable metric. "Trial-to-paid conversion rate" is actionable because the product team can redesign the trial experience, the sales team can adjust outreach timing, and the pricing team can test different plan structures. If no team can draw a direct line from their work to the metric, it belongs higher in the tree, not at the leaf level.

3. 3

   Favour rates over absolute numbers

   Absolute numbers like "10,000 new sign-ups" conflate volume with quality. Rates like "visitor-to-sign-up conversion rate" isolate the efficiency of a specific transition and are more diagnostic. When a rate drops, you know the process is degrading regardless of volume. When an absolute number drops, you cannot tell whether the problem is fewer inputs or a worse conversion. Use absolute numbers at the top of the tree for context, but prefer rates at the operational level.

4. 4

   Match the metric to your growth stage

   Pre-product-market-fit companies should focus almost exclusively on Activation and Retention. If users are not experiencing value and coming back, pouring resources into Acquisition is wasteful. Post-product-market-fit companies can expand focus to Acquisition channels and Revenue optimisation. Mature companies with strong Retention can invest in Referral loops. The metric tree should reflect where you are, not where you aspire to be.

5. 5

   Validate with cohort analysis

   Once you have chosen metrics, validate them by running cohort analysis. Do users with higher Activation scores retain better? Do retained users with deeper feature adoption generate more Revenue? Do high-NPS users actually refer more? If the answer is no, the metric you chose for that stage is not capturing what matters and needs to be replaced. Cohort analysis is the empirical check that prevents your tree from being built on assumptions.

A useful pattern is to start with one primary metric per stage and then decompose each into two or three sub-metrics as your understanding deepens. For Acquisition, start with "new sign-ups per week" and then decompose into sign-ups by channel, cost per sign-up by channel, and sign-up-to-activation rate by channel. For Activation, start with "percentage of new users who complete the key action within seven days" and decompose into each step of the onboarding flow. This progressive decomposition keeps the tree manageable while preserving the depth you need for diagnosis.

Every growth-stage company has a bottleneck: the single stage in the AARRR framework that, if improved, would have the largest impact on the top-level outcome. The challenge is that flat dashboards make every underperforming stage look equally urgent. Acquisition is down, Activation is flat, Retention is slipping. Where do you focus? Without a metric tree, this question devolves into a debate between the loudest voices in the room. With a metric tree, you can answer it structurally.

Start at the root of your tree and walk downward. At each level, ask: which child node has the weakest performance relative to its potential? If your Acquisition volume is healthy but your Activation rate is 15% when benchmarks suggest 30-40%, Activation is likely the bottleneck. But do not stop there. Open the Activation branch and look at the sub-drivers. Is onboarding completion the problem, or are users completing onboarding but failing to reach the key action? Is time to first value too long, or is the key action itself unclear? The bottleneck is rarely at the stage level. It is at a specific node two or three levels down.

This process of walking the tree from root to leaf is what separates metric-tree-driven growth teams from dashboard-driven ones. A dashboard team sees "Activation is 15%" and launches a general onboarding improvement initiative. A tree-driven team sees that onboarding completion is actually 78%, but only 19% of users who complete onboarding perform the key action within seven days. The bottleneck is not onboarding itself; it is the gap between completing onboarding and reaching the aha moment. That is a much more specific, much more solvable problem.

The AARRR framework gave startups a shared vocabulary for growth. Metric trees give that vocabulary a structure. A flat funnel tells you that Retention is your weakest stage. A metric tree tells you that Retention is weak because feature adoption depth among users acquired through paid channels is 40% lower than organic users, and the root cause is that paid users land on a generic onboarding flow that does not match the intent signal from the ad they clicked. That level of specificity is the difference between a quarterly initiative and a two-week fix that moves the number. When you combine the intuitive clarity of AARRR with the causal depth of a metric tree, you get a growth system that is both easy to communicate and powerful enough to act on.