mittuled funnel-analyser-growth
Agent: Growth Engineer
L2 growth engineer (Nx) responsible for growth instrumentation, metrics dashboards, funnel analysis, and growth loop activation.
Department ethos: ideal-data-growth.md Tool policy: allowed-tools.yaml
Skill Description
The growth funnel analyser examines the acquisition-to-activation funnel — from first touch through signup, onboarding, and activation — to identify the step with the largest absolute drop-off, segment by channel and cohort, and recommend experiments that maximize conversion through the growth funnel.
When to Use
- When acquisition costs rise without corresponding growth in activated users.
- When a growth experiment needs baseline conversion data across funnel steps.
- When a new onboarding flow ships and the team needs to measure its impact on activation rates.
- When the growth lead requests a funnel audit to prioritize the next experiment cycle.
Workflow
- Define growth funnel steps: Map the funnel from first touch (ad impression, organic visit) through landing page, signup, onboarding, and activation. Confirm each step has a corresponding tracking event.
- Pull conversion data: Query step-to-step conversion rates for the analysis window. Compute both step-to-step and end-to-end conversion rates.
- Rank drop-off points: Identify the step with the largest absolute user loss. Calculate the potential user gain if that step's conversion improved by 10-20%.
- Segment by channel: Break the funnel by acquisition channel (paid search, organic, referral, direct). Identify channels with systematically worse conversion at specific steps.
- Segment by device and cohort: Compare conversion across mobile vs. desktop and by signup cohort (weekly). Flag device-specific UX issues and temporal trends.
- Produce recommendations: For the top 2 drop-off points, hypothesize root causes and propose experiments with expected conversion lift. Prioritize by impact multiplied by confidence.
Anti-Patterns
- Optimizing the wrong step: Targeting the step with the lowest conversion rate rather than the step with the largest absolute drop-off wastes experiment capacity. Why: a 50% to 45% drop with 100K users entering is higher-leverage than a 10% to 5% drop with 1K users entering.
- Channel-blind analysis: Reporting one funnel for all channels hides that organic users may convert at 3x the rate of paid users. Why: the bottleneck step differs by channel; a single funnel view prescribes a generic intervention.
- No baseline before experiments: Launching a funnel experiment without a stable baseline makes it impossible to attribute lift. Why: if the baseline is noisy, the experiment's signal is indistinguishable from variance.
Output
Success:
- A growth funnel report with step-by-step conversion rates, drop-off rankings, channel and device segmentation, and 2-3 experiment recommendations with projected impact.
Failure:
- Growth funnel events are missing or inconsistent, producing gaps in the analysis. Report the missing events and the instrumentation work required before reliable analysis is possible.
Related Skills
funnel-analyser-- the data analyst's funnel analyser covers the full product funnel; this skill focuses specifically on acquisition-to-activation.growth-loop-optimiser-- funnel analysis within growth loops feeds loop optimisation decisions.activation-moment-validator-- the activation step in the funnel must be validated as a retention predictor.metrics-dashboard-growth-- growth funnel metrics are displayed on the growth dashboard.