Google Analytics 4 Funnel Exploration Setup
Funnels Are a Reporting Lens, Not a New Data Model
GA4's Funnel Exploration doesn't require any special event setup beyond the events you'd already be tracking — it's a reporting technique that takes an ordered sequence of existing events and shows drop-off between each step. The configuration decisions that matter are: open vs closed funnel, and the step condition logic, not the underlying data collection.
Open vs Closed Funnels
A closed funnel requires users to enter at step 1 before being counted at all — anyone who skips straight to step 2 (e.g. arrives directly on a checkout page from a saved link) is excluded. An open funnel counts users regardless of entry point, showing step 2 completion even if they never completed step 1. Closed funnels are stricter and better for understanding a designed, sequential flow (e.g. a checkout process); open funnels are better for understanding overall engagement with a set of milestones users might reach in any order.
{
"funnel_type": "closed",
"steps": [
{ "name": "View Item", "event": "view_item" },
{ "name": "Add to Cart", "event": "add_to_cart" },
{
"name": "Begin Checkout",
"event": "begin_checkout",
"condition": { "param": "value", "operator": ">", "value": 0 }
},
{ "name": "Add Shipping Info", "event": "add_shipping_info" },
{ "name": "Purchase", "event": "purchase" }
],
"elapsed_time_shown": true,
"breakdown_dimension": "device_category"
}Elapsed Time Reveals Friction Even Without Drop-Off
Breakdowns Turn a Funnel Into a Diagnostic Tool
Applying a breakdown dimension (device category, traffic source/medium, new vs returning user, or a custom user property) to a funnel splits each step's completion rate by that dimension, which is usually where the actionable insight actually lives. A 40% drop-off at Add Shipping Info might be perfectly normal in aggregate but mask a 70% drop-off specifically on mobile Safari — invisible until you break it down.
Common Setup Mistakes
The most frequent error is using an event that fires too often or too loosely as a funnel step — e.g. using a generic page_view with a URL condition instead of a purpose-built custom event, which makes step conditions brittle against URL parameter changes or redirects. Another common issue: forgetting that GA4 funnels respect your reporting identity settings (Blended, Observed, etc.), which changes how cross-device users are stitched together and can materially shift completion rates between identity models.
Sampling Applies at High Volume
What's Next?
Funnels show *where* people drop off, but not the *paths* they took to get there or away from the intended flow. Path Exploration is the natural next analytical layer for understanding the branching behavior funnels can't capture.
I build these systems professionally.
Whether it's a RAG pipeline, analytics migration, or AI workflow — let's talk.