You can measure two hundred things in your product. Impressions, sessions, CTR, bounce rate, scroll depth… and still not know why you are not growing. The AARRR funnel exists for the opposite: five stages, a handful of metrics, and one clear question — where is your growth leaking?
It was created by Dave McClure (500 Startups) and is known as “pirate metrics” because of the acronym: Acquisition, Activation, Retention, Revenue, Referral. Almost two decades later, it is still the most useful framework for diagnosing a digital business. In this guide we bring it down to earth: what to measure at each stage, indicative benchmarks, and the mistakes we see over and over again.
The five stages at a glance
| Stage | Question | Typical metrics |
|---|---|---|
| Acquisition | How do people discover you? | Traffic by channel, CAC, visit→sign-up conversion |
| Activation | Do they perceive value quickly? | % completing the “aha moment”, time-to-value |
| Retention | Do they come back? | D7/D30 retention, churn, DAU/MAU |
| Revenue | Do they pay? | Conversion to paid, ARPU, LTV, LTV/CAC |
| Referral | Do they recommend you? | NPS, viral coefficient (k), invitations sent |
Acquisition: not all channels are equal
The question is not “how much traffic do I have?” but “which channel brings me users who stick around, and at what cost?”
What to measure:
- Traffic and conversion by channel (organic, paid, direct, referral, email), not in aggregate.
- CAC per channel: the cost of acquiring a customer, not a click.
- Downstream quality: D30 retention of the users from each channel. It is common to discover that the cheapest channel in CPC terms is the most expensive in real customers.
The classic mistake: optimizing for volume. A channel that brings 10,000 visits with a 0.2% sign-up rate is worth less than one that brings 1,000 with 8%. And the opposite bias exists too: killing SEO because “it takes too long”, when it is the channel that compounds over time — it is the foundation of cases like trasterOne, which positioned itself as the number one comparison site in Spain with programmatic SEO.
Activation: the most underrated stage
A sign-up is not a user. Activation is the first time someone experiences the value of your product: the “aha moment”. In Slack it is sending messages with your team; in a comparison site, seeing relevant results in your area; in a B2B SaaS, completing the first integration.
What to measure:
- % of sign-ups that complete the value action (your definition of activation).
- Time-to-value: how long it takes them to get there. Minutes and days are different worlds.
- Step-by-step onboarding funnel, to see the exact drop-off point.
If you do not know what your “aha moment” is, find it with data: analyze which early actions separate the users who are still active a month later from those who disappear. That correlation is your candidate activation metric.
Retention: the metric that predicts the future
Everything else can be going well; if retention fails, the business is a leaky bucket. Retention is the best test of product-market fit and the silent multiplier of everything else: doubling retention is usually worth more than doubling traffic.
What to measure:
- Retention curves by cohort (D1/D7/D30 in consumer; weeks or months in B2B). What matters is not the isolated number, but whether the curve flattens out — a sign that a core of users found lasting value.
- Monthly churn (customers and revenue) in subscription businesses.
- DAU/MAU as a proxy for usage frequency, if your product is meant for recurring use.
Classic mistake: looking only at new users. A dashboard with growing acquisition can hide cohorts that empty out within a month. Always cohorts, never aggregates.
Revenue: monetizing without breaking the rest of the funnel
What to measure:
- Free→paid conversion (or visit→purchase in ecommerce).
- ARPU and average order value.
- LTV (customer lifetime value) and the LTV/CAC ratio — the metric that sums up the economic health of your growth. The classic reference: LTV/CAC ≥ 3 and recovering CAC in under 12 months.
Revenue is rarely fixed at the revenue stage: if conversion to paid is low, the cause usually lies in activation (users did not perceive value) or in pricing (it is not aligned with perceived value). Experimenting with pricing — packages, anchors, trials — is among the most profitable work in growth and among the least practiced.
Referral: the multiplier (if the product holds up)
What to measure:
- NPS as a thermometer (who would recommend you?).
- Viral coefficient (k): invitations sent × conversion rate of those invitations. With k=0.5, every 10 customers bring 5 for free: it is not virality, but it cuts your effective CAC in half.
- % of sign-ups from invitations or word of mouth (ask in onboarding: “how did you hear about us?”).
Golden rule: do not build a referral program on top of a product with poor retention. You would be amplifying disappointment. Referral is the last stage you optimize, not the first.
How to use the funnel to prioritize (without drowning in metrics)
- Instrument one event per stage. Five well-defined events are worth more than fifty badly placed ones. With GA4, PostHog, or Mixpanel it is an afternoon’s work… if the tracking is set up properly.
- Build the full funnel and calculate the ratios between stages.
- Find the bottleneck: the biggest drop compared to benchmarks in your sector.
- Concentrate the next cycle’s experiments there. A funnel is optimized stage by stage, not all at once.
- Repeat. When you unblock one stage, the bottleneck moves to another. That is a good sign: it means you are growing.
And one important nuance: the AARRR funnel is a diagnostic model, not a literal description of human behavior. Users enter, leave, and skip stages. To design sustainable growth mechanics, the funnel is complemented with growth loops — loops where the output of one stage feeds an earlier one (content that generates SEO that generates users who generate content). We will cover it in detail in an upcoming article.
Conclusion
The value of AARRR is not in the metrics: it is in the conversation it forces. You stop arguing opinions (“we need more brand”, “we should do TikTok”) and start arguing data (“we lose 70% between sign-up and activation; let’s fix that first”).
Five stages, one event per stage, cohorts, and one bottleneck per quarter. It is less glamorous than a growth hack — and it works considerably better.
Not sure which stage your growth is leaking from? We will tell you with a funnel audit →