GA4 Funnel Analysis

Process

1. Verify auth: python scripts/ga4_auth.py --check
2. If the user did not specify --steps or --preset, propose a funnel based on what events fire on the property and ask for confirmation.
3. Validate event coverage with the events script: python scripts/ga4_events.py --property <id> --check-events <step1,step2,...> --days 7
4. Spawn the ga4-funnel agent with the property ID and validated step list.
5. Return agent output with an optional follow-up offer.

Funnel inputs

• --steps event1,event2,event3 - arbitrary list of GA4 event names; the funnel runs in the listed order.
• --preset ecomm - convenience preset for the recommended e-commerce purchase funnel: view_item -> add_to_cart -> begin_checkout -> add_payment_info -> purchase.
• Neither flag: the agent proposes a funnel based on which events fire on the property.

Optional checks

• --check-postpayment (off by default): runs the post-payment heuristic against add_payment_info. If the count is within 10% of purchase, the result is a hypothesis to investigate; it neither proves order nor removes a step.
• --breakdown <dim> - funnel split by one dimension (e.g. deviceCategory, sessionDefaultChannelGroup).

Date range

Default 28 days. Override with --days N.

After analysis

Offer: "Want to break down the drop-off by segment? Use /ga4 segments <property-id>"

Funnel reports accept 2-10 unique steps and retain missing steps by default. --drop-missing explicitly changes the requested funnel. --check-postpayment is advisory and never removes steps. Validation uses the complete requested date window. Numbered API step names are normalized; breakdown rates use the named dimension and RESERVED_TOTAL rows, never summed segment populations. Missing totals or malformed rows produce an error instead of fabricated rates.