activation-funnel

What this skill does

# Activation Funnel Expert

## Overview

A funnel is the single most useful diagnostic tool a growth or onboarding PM owns. It turns a fuzzy product story ("users drop off somewhere") into a numbered, actionable picture ("76% land, 41% start setup, 9% finish setup, 4% take the activation action -- the biggest drop is between start and finish setup at 32 percentage points").

This skill specifies funnel structures using **Dave McClure's AARRR** (Acquisition, Activation, Retention, Revenue, Referral) and its broader cousin **AAARRR** (which adds Awareness on the front), and analyzes them with a stdlib Python tool (`funnel_analyzer.py`). The tool ingests a JSON funnel definition (stages with counts), and outputs (a) stage-by-stage conversion and drop-off, (b) a Mermaid flowchart with conversion percentages on each arrow, (c) a bottleneck call-out for the largest absolute drop. It supports all six SHARED_OUTPUT_SCHEMA formats so the analysis travels into Jira, Linear, Confluence, Notion, or a PR.

The activation step is the centerpiece. Sean Ellis defined "activated user" as a user who has done the thing that, statistically, predicts retention. Slack's classic 2000-messages-per-team activation event, Facebook's "7 friends in 10 days", and Dropbox's "1 file in 1 folder on 1 device" are all activation definitions, not retention metrics. Pin the activation event before you optimize the funnel that leads to it.

### When to Use

- **New onboarding design** -- you are designing the user journey from signup to value and need to specify the funnel events.
- **Drop-off diagnosis** -- onboarding completion is flat but signups grow; where is the leak?
- **Activation metric definition** -- the team needs to pick the "aha" event that predicts long-term retention.
- **Cross-functional alignment** -- marketing optimizes acquisition, product optimizes activation, neither agrees on the funnel definition.
- **Quarterly business review** -- one canonical funnel chart drives the QBR conversation about growth.
- **A/B test design** -- a funnel makes the experiment's primary and counter-metric obvious.

### When NOT to Use

- For pre-product-market-fit discovery (use `discovery/brainstorm-ideas/`, `discovery/identify-assumptions/`; funnels presume a stable user journey).
- For pure top-of-funnel marketing-channel attribution (use marketing-side tools).
- For revenue-cohort retention analysis specifically (use the data-analytics domain; funnels capture single-flow conversion, cohorts capture time-based retention).
- When the team has not instrumented the events -- a funnel without telemetry is wishful thinking.

## Pirate Metrics framework (AARRR)

Dave McClure proposed Pirate Metrics in 2007 as a five-stage growth funnel:

| Stage | What it measures | Typical events |
|---|---|---|
| **Acquisition** | Users find you | Landing-page visit, ad click, organic search visit |
| **Activation** | First good experience | Signup, profile complete, "aha" event |
| **Retention** | They come back | D1, D7, D30 active; weekly active |
| **Revenue** | They pay | Trial-to-paid, upgrade, expansion |
| **Referral** | They tell others | Invite sent, invite accepted, share rate |

McClure's insight was that most teams obsess over Acquisition (the top) and Revenue (the bottom), while the leverage is usually in Activation and Retention. A 5-point improvement in activation flows through to retention, revenue, and referral; a 5-point improvement in acquisition often gets eaten by churn.

### AAARRR (Awareness-prefixed)

Some teams add **Awareness** at the front: how many people know your product exists at all? AAARRR is useful when the brand is new or the category is undefined; AARRR is fine when product-market fit is established.

| Stage | Question |
|---|---|
| **Awareness** | Do they know you exist? (impressions, mentions, branded search) |
| **Acquisition** | Do they visit you? (visits, signups-started) |
| **Activation** | Do they get value? (the aha event) |
| **Retention** | Do they come back? (D1, D7, D30) |
| **Revenue** | Do they pay? (trial-to-paid) |
| **Referral** | Do they bring friends? (invites, shares) |

## Defining activation: the "aha" event

Sean Ellis introduced the **activation-rate framework**: the activation event is the action that, statistically, separates users who retain from users who churn. Find it by:

1. Identify the cohort of users who are still active at D30 (or your retention horizon).
2. Look at what they did in their first session / first week that the churned cohort did not do.
3. The behavior with the highest correlation to D30 retention is the activation event candidate.

### Real activation events (publicly discussed)

| Product | Activation event |
|---|---|
| Slack | A team sends 2,000 messages |
| Facebook | A user adds 7 friends in 10 days |
| Dropbox | A user puts 1 file in 1 folder on 1 device |
| Airbnb (host) | A host books their first guest |
| Airbnb (guest) | A guest books their first trip |
| Notion | A user creates 3+ pages and invites 1+ collaborator |
| Spotify | A user listens to 25+ minutes in their first week |
| Duolingo | A user completes 1 lesson in 3 of their first 7 days |
| Twitter | A user follows 30+ accounts |
| HubSpot | A user adds 5 contacts and 2 deals |
| Pinterest | A user follows 5 boards and saves 1 pin |
| LinkedIn | A user adds 5 connections in their first week |
| Figma | A user creates a file and adds 2+ collaborators |
| Canva | A user creates a design and downloads or shares it |

The pattern: a **specific count + a specific time window + a specific action**. Vague "completed onboarding" is not an activation event -- it is an output. A good activation event is observable, instrumented, and predictive of long-term value.

## Funnel math

### Conversion rate

For each stage transition: `conversion[i] = users[i+1] / users[i]`

### Drop-off (absolute)

`drop[i] = users[i] - users[i+1]`

### Drop-off (rate)

`drop_rate[i] = 1 - conversion[i]`

### Cumulative conversion

`cumulative[i] = users[i] / users[0]`

The largest absolute drop is where the funnel leaks the most users. The largest *relative* drop (worst conversion rate) is where the stage is most broken. Both signals matter; this skill's tool surfaces both.

## Counter-metric pairing

Funnel optimization is famously easy to game. If you measure only conversion, you can boost it by lowering the bar at each stage:

- Drop email verification -> signup conversion goes up; fake-account problem goes up too.
- Auto-import data -> "activation" rate climbs; users feel nothing happened.
- Add a "skip" button -> setup completion climbs; D7 retention drops.

For every funnel stage, declare a **counter-metric** that must not degrade. The pair (conversion, counter-metric) is what gets reviewed -- not conversion alone.

| Funnel stage | Primary metric | Counter-metric |
|---|---|---|
| Acquisition | Signups / visitors | Quality signup rate (e.g., real-email %) |
| Activation | Activation rate | D7 retention of activated users |
| Retention | D30 retention | Engagement depth (events / active user) |
| Revenue | Trial-to-paid | First-month churn |
| Referral | Invite acceptance | Quality of referred user (do they activate?) |

## Leading vs lagging activation indicators

| Indicator | Timing | Use |
|---|---|---|
| **Leading** | Hours / days into use | Daily dashboard, on-call signal (e.g., "step 2 completion in first session") |
| **Activation** | First N days | The headline activation rate |
| **Lagging** | D30 retention of activated users | Quarterly truth check (is the activation event still predictive?) |

Leading indicators move first, so they are what the team optimizes. The activation rate is the headline. The lagging indicator audits whether the activation definition is still correct.

## Workflow

1. **Pick the framework.** AARRR for established product, AAARRR if the category is undefined.
2. **Define each stage as an event.** Each transition is `event A -> event B`. N