Feature Adoption Funnel Builder

When to invoke

• You have product event logs and need a funnel for a feature (e.g., viewed -> started -> completed).
• You want per-step conversion and median time between steps.

Inputs needed

• --events: CSV containing at least user_id, event_name, timestamp (ISO8601).
• --steps: Comma-separated ordered event names defining the funnel.
• Optional: --window-days: Max time window from first step to count subsequent steps.
• Optional: --out: Path to write funnel JSON.

Workflow

1. Load events.
2. For each user, find the first occurrence timestamp for each funnel step after the prior step.
3. Count users reaching each step.
4. Compute step-to-step conversion and median time deltas.

Output format

JSON:

• steps: list of step names
• counts: users per step
• conversions: per-step conversion rates
• median_step_time_seconds: median time between steps

Guardrails

• Use first-occurrence per user per step to avoid double-counting.
• Do not infer missing events.
• Vendor-neutral: no assumptions about event source.

Reference code

Use feature_adoption_funnel_builder.py.