Funnel Analyser

funnel-analyser

Agent: Data Analyst

L2 data analyst (Nx) responsible for data modelling, instrumentation implementation, metrics dashboards, funnel analysis, and signal synthesis.

Department ethos: ideal-data-growth.md

Skill Description

The funnel analyser diagnoses conversion funnels by computing step-to-step conversion rates, identifying the highest-impact drop-off points, segmenting by cohort to isolate root causes, and producing prioritized recommendations for improving end-to-end conversion.

When to Use

• When a conversion metric (signup-to-activation, trial-to-paid, checkout completion) declines and the team needs to locate the drop-off.
• When a new user flow launches and needs baseline funnel measurement.
• When the growth team prioritizes experiments and needs to identify the highest-leverage funnel step to target.
• When a product change ships and the team needs to measure its impact on funnel conversion.

Workflow

1. Define funnel steps: Enumerate each step in the conversion flow with the corresponding event name. Confirm instrumentation coverage for every step.
2. Pull conversion data: Query the analytics warehouse for event counts at each step over the analysis window (typically 7-30 days). Compute step-to-step and end-to-end conversion rates.
3. Identify drop-off points: Rank steps by absolute drop-off volume and relative conversion rate. Highlight the step with the largest gap between entry and exit.
4. Segment by cohort: Break the funnel by user segment (new vs. returning, acquisition channel, device, plan tier). Identify segments with significantly different conversion patterns.
5. Time-based analysis: Plot conversion rates over time to detect trends, seasonality, or regressions coinciding with product releases.
6. Produce recommendations: For the top 2-3 drop-off points, hypothesize root causes (UX friction, unclear copy, missing trust signals, technical errors) and propose experiments to address each.

Anti-Patterns

• Aggregate-only analysis: Reporting a single conversion rate without segmentation hides that one segment may convert at 80% while another converts at 5%. Why: the intervention differs by segment; aggregate data obscures this.
• Ignoring time-to-convert: Measuring only whether users convert, not how long it takes, misses latency-driven abandonment. Why: a step that takes 3 minutes on mobile but 30 seconds on desktop reveals a platform-specific UX issue.
• No event verification: Analysing a funnel without first confirming that every step event fires correctly risks drawing conclusions from broken data. Why: a missing event looks like a 100% drop-off, triggering a false alarm.

Output

Success:

• A funnel analysis report containing step-by-step conversion rates, drop-off rankings, cohort segmentation, time-series trends, and prioritized recommendations with hypothesized root causes.

Failure:

• One or more funnel steps lack instrumentation, making the analysis incomplete. Report the missing events, the data gap, and the instrumentation work required.

Related Skills

• adoption-tracker-data -- adoption tracking uses funnel analysis to identify where feature discovery and trial break down.
• metrics-dashboard-builder -- funnel metrics are surfaced on dashboards for ongoing monitoring.
• funnel-analyser-growth -- the growth engineer's funnel analyser focuses on acquisition and activation; this skill covers the full product funnel.
• statistical-significance-tracker -- experiments targeting funnel steps need significance tracking to validate results.