Building a GA4 Funnel Exploration to Find the Exact Step Where People Leave
Every business with a process that has more than one step — a quote request, a booking calendar, a checkout — has a point where people quietly give up, and most owners have no real idea where. They look at one number ("3% of visitors convert"), decide the whole site is broken, and redesign the homepage. Three weeks later nothing's moved, because the actual leak was two steps downstream, on a form field nobody thought to check. 💖 We'd rather spend twenty minutes building a proper funnel than another twenty hours guessing.
What most businesses get wrong
The mistake is treating "conversion rate" as one health metric instead of a sequence of separate decisions. A 3% overall rate could mean "most people get to the final step and bail at payment" or "90% never get past step one" — two completely different problems with two different fixes, and you can't tell them apart from a top-line number. GA4's standard reports are built around sessions and events, not sequences, so you have to deliberately build the Exploration to see each step's drop-off as its own percentage.
- Check your events first. In GA4 Admin > Events (or DebugView), confirm the events you're about to use as steps are actually firing. A funnel built on broken tracking just tells you a confident lie.
- Explore > blank > Funnel Exploration technique, named clearly so you can find it again (e.g. "Quote Funnel — Oct 2026").
- Define 3-6 steps using real events or page path conditions — e.g. Step 1: page_view on /services, Step 2: a view or scroll event, Step 3: form_start, Step 4: generate_lead or purchase.
- Choose open vs closed funnel. Closed only counts people who hit every step in exact order; open is more forgiving and usually closer to reality.
- Turn on "elapsed time" to see how long people sit at each step — a slow step is often a confusing one.
- Add a breakdown dimension — device category or channel group — and watch the chart split.
How to actually read what you've built
The step chart is the headline, but the "trend over time" tab shows whether a step's drop-off is worsening over weeks — how you catch a regression after a site update. The breakdown dimension is where the real insight lives — a 50% aggregate drop might be 20% on desktop and 85% on mobile, and averaging them hides the whole story. Elapsed time tells you whether people leave instantly (confusion, a broken element) or linger first (genuinely weighing it up, or struggling).
Mistakes to avoid
- Building the funnel on page views alone when the real moment is a click or scroll event with no page load.
- Reading the aggregate conversion rate and stopping there, instead of checking where the drop actually happens.
- Never adding a breakdown dimension — the aggregate number usually hides a more interesting split.
- Building the funnel once and never revisiting — seasonality and site updates shift where the leak sits.
- Assuming a near-zero step means "nobody does this" rather than checking if the event stopped firing.
Frequently asked questions
How many steps should a funnel have?
Three to six for most small-business processes. Fewer rarely tells you anything new; more gets noisy.
Do I need special access to build this in GA4?
Analyst-level access to build and save one; Viewer access is enough to look at one that's shared with you.
Will this tell me why people are leaving at a particular step?
Honestly, no — the main limit worth naming. It shows where, not why. For that you still need something qualitative alongside it, like session recordings or simply testing the step yourself on the device where the drop is worst.
Can I use this for an online checkout rather than a lead form?
Yes — if your GA4 ecommerce events are set up (view_item, add_to_cart, begin_checkout, purchase), they map almost directly onto funnel steps.
Keep reading 🤍
I help Gold Coast and Brisbane businesses grow with branding, websites and marketing that actually works.
Work with me ✦